ift6266: writeup/mlj_submission/aigaion-shorter.bib annotate

annotate writeup/mlj_submission/aigaion-shorter.bib @ 646:220f5ab44457

adding license

author	Razvan Pascanu <r.pascanu@gmail.com>
date	Wed, 17 Oct 2012 09:16:32 -0400
parents	b1be957dd1be
children

rev	line source
587 b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1 %Aigaion2 BibTeX export from LISA - Publications
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2 %Tuesday 01 June 2010 10:46:52 AM
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3 @INPROCEEDINGS{Attardi+al-2009,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4 author = {Attardi, Giuseppe and Dell'Orletta, Felice and Simi, Maria and Turian, Joseph},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	5 keywords = {classifier, dependency parsing, natural language, parser, perceptron},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	6 title = {Accurate Dependency Parsing with a Stacked Multilayer Perceptron},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	7 booktitle = {Proceeding of Evalita 2009},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	8 series = {LNCS},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	9 year = {2009},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	10 publisher = {Springer},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	11 abstract = {Abstract. DeSR is a statistical transition-based dependency parser which learns from annotated corpora which actions to perform for building parse trees while scanning a sentence. We describe recent improvements to the parser, in particular stacked parsing, exploiting a beam search strategy and using a Multilayer Perceptron classifier. For the Evalita 2009 Dependency Parsing task DesR was configured to use a combination of stacked parsers. The stacked combination achieved the best accuracy scores in both the main and pilot subtasks. The contribution to the result of various choices is analyzed, in particular for taking advantage of the peculiar features of the TUT Treebank.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	12 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	13
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	14 @INPROCEEDINGS{Bengio+al-2009,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	15 author = {Bengio, Yoshua and Louradour, Jerome and Collobert, Ronan and Weston, Jason},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	16 title = {Curriculum Learning},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	17 year = {2009},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	18 crossref = {ICML09-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	19 abstract = {Humans and animals learn much better when the examples are not randomly presented but organized in a meaningful order which illustrates gradually more concepts, and more complex ones. Here, we formalize such training strategies in the context of machine learning, and call them 'curriculum learning'. In the context of recent research studying the difficulty of training in the presence of non-convex training criteria (for deep deterministic and stochastic neural networks), we explore curriculum learning in various set-ups. The experiments show that significant improvements in generalization can be achieved by using a particular curriculum, i.e., the selection and order of training examples. We hypothesize that curriculum learning has both an effect on the speed of convergence of the training process to a minimum and, in the case of non-convex criteria, on the quality of the local minima obtained: curriculum learning can be seen as a particular form of continuation method (a general strategy for global optimization of non-convex functions).}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	20 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	21
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	22 @TECHREPORT{Bengio+al-2009-TR,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	23 author = {Bengio, Yoshua and Louradour, Jerome and Collobert, Ronan and Weston, Jason},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	24 title = {Curriculum Learning},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	25 number = {1330},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	26 year = {2009},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	27 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	28 abstract = {Humans and animals learn much better when the examples are not randomly presented but organized in a meaningful order which illustrates gradually more concepts, and gradually more complex ones. Here, we formalize such training strategies in the context of machine learning, and call them 'curriculum learning'. In the context of recent research studying the difﬁculty of training in the presence of non-convex training criteria (for deep deterministic and stochastic neural networks), we explore curriculum learning in various set-ups. The experiments show that signiﬁcant improvements in generalization can be achieved. We hypothesize that curriculum learning has both an effect on the speed of convergence of the training process to a minimum and, in the case of non-convex criteria, on the quality of the local minima obtained: curriculum learning can be seen as a particular form of continuation method (a general strategy for global optimization of non-convex functions).}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	29 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	30
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	31 @MISC{Bengio+al-patent-2000,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	32 author = {Bengio, Yoshua and Bottou, {L{\'{e}}on} and {LeCun}, Yann},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	33 title = {Module for constructing trainable modular network in which each module outputs and inputs data structured as a graph},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	34 year = {2000},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	35 howpublished = {U.S. Patent 6,128,606, October 3}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	36 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	37
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	38 @MISC{Bengio+al-patent-2001,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	39 author = {Bengio, Yoshua and Bottou, {L{\'{e}}on} and G. Howard, Paul},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	40 title = {Z-Coder : a fast adaptive binary arithmetic coder},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	41 year = {2001},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	42 howpublished = {U.S. Patent 6,188,334, February 13, 2001, along with patents 6,225,925, 6,281,817, and 6,476,740}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	43 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	44
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	45 @MISC{Bengio+al-patent-94,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	46 author = {Bengio, Yoshua and {LeCun}, Yann and Nohl, Craig and Burges, Chris},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	47 title = {Visitor Registration System Using Automatic Handwriting Recognition},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	48 year = {1994},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	49 howpublished = {Patent submitted in the U.S.A. in October 1994, submission number 1-16-18-1}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	50 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	51
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	52 @INCOLLECTION{Bengio+al-spectral-2006,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	53 author = {Bengio, Yoshua and Delalleau, Olivier and Le Roux, Nicolas and Paiement, Jean-Fran{\c c}ois and Vincent, Pascal and Ouimet, Marie},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	54 editor = {Guyon, Isabelle and Gunn, Steve and Nikravesh, Masoud and Zadeh, Lofti},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	55 title = {Spectral Dimensionality Reduction},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	56 booktitle = {Feature Extraction, Foundations and Applications},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	57 year = {2006},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	58 publisher = {Springer},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	59 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/eigenfn_chapter.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	60 abstract = {In this chapter, we study and put under a common framework a number
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	61 of non-linear dimensionality reduction methods, such as Locally Linear Embedding,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	62 Isomap, Laplacian eigenmaps and kernel {PCA}, which are based
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	63 on performing an eigen-decomposition (hence the name "spectral"). That
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	64 framework also includes classical methods such as {PCA} and metric multidimensional
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	65 scaling ({MDS}). It also includes the data transformation step used
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	66 in spectral clustering. We show that in all of these cases the learning algorithm
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	67 estimates the principal eigenfunctions of an operator that depends on
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	68 the unknown data density and on a kernel that is not necessarily positive
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	69 semi-definite. This helps to generalize some of these algorithms so as to predict
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	70 an embedding for out-of-sample examples without having to retrain the
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	71 model. It also makes it more transparent what these algorithm are minimizing
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	72 on the empirical data and gives a corresponding notion of generalization
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	73 error.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	74 cat={B},topics={HighDimensional,Kernel,Unsupervised},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	75 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	76
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	77 @INCOLLECTION{Bengio+al-ssl-2006,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	78 author = {Bengio, Yoshua and Delalleau, Olivier and Le Roux, Nicolas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	79 editor = {Chapelle, Olivier and {Sch{\"{o}}lkopf}, Bernhard and Zien, Alexander},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	80 title = {Label Propagation and Quadratic Criterion},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	81 booktitle = {Semi-Supervised Learning},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	82 year = {2006},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	83 pages = {193--216},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	84 publisher = {{MIT} Press},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	85 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/bengio_ssl.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	86 abstract = {Various graph-based algorithms for semi-supervised learning have been proposed in
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	87 the recent literature. They rely on the idea of building a graph whose nodes are
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	88 data points (labeled and unlabeled) and edges represent similarities between points.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	89 Known labels are used to propagate information through the graph in order to label
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	90 all nodes. In this chapter, we show how these different algorithms can be cast into
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	91 a common framework where one minimizes a quadratic cost criterion whose closed-form solution is found by solving a linear system of size n (total number of data
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	92 points). The cost criterion naturally leads to an extension of such algorithms to
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	93 the inductive setting, where one obtains test samples one at a time: the derived
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	94 induction formula can be evaluated in O(n) time, which is much more efficient
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	95 than solving again exactly the linear system (which in general costs O(kn2) time
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	96 for a sparse graph where each data point has k neighbors). We also use this inductive
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	97 formula to show that when the similarity between points satisfies a locality property,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	98 then the algorithms are plagued by the curse of dimensionality, with respect to the
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	99 dimensionality of an underlying manifold.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	100 cat={B},topics={Unsupervised},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	101 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	102
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	103 @TECHREPORT{Bengio+al-treecurse-2007,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	104 author = {Bengio, Yoshua and Delalleau, Olivier and Simard, Clarence},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	105 title = {Decision Trees do not Generalize to New Variations},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	106 number = {1304},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	107 year = {2007},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	108 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	109 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/bengio+al-tr1304.pdf}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	110 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	111
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	112 @INPROCEEDINGS{Bengio+Bengio96,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	113 author = {Bengio, Samy and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	114 editor = {Xu, L.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	115 title = {An {EM} Algorithm for Asynchronous Input/Output Hidden {M}arkov Models},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	116 booktitle = {International Conference On Neural Information Processing},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	117 year = {1996},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	118 pages = {328--334},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	119 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/iconip96.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	120 abstract = {In learning tasks in which input sequences are mapped to output sequences, it is often the case that the input and output sequences are not synchronous. For example, in speech recognition, acoustic sequences are longer than phoneme sequences. Input/Output Hidden {Markov} Models have already been proposed to represent the distribution of an output sequence given an input sequence of the same length. We extend here this model to the case of asynchronous sequences_ and show an Expectation-Maximization algorithm for training such models.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	121 topics={Markov},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	122 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	123
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	124 @INCOLLECTION{Bengio+chapter2007,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	125 author = {Bengio, Yoshua and {LeCun}, Yann},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	126 editor = {Bottou, {L{\'{e}}on} and Chapelle, Olivier and DeCoste, D. and Weston, J.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	127 title = {Scaling Learning Algorithms towards {AI}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	128 booktitle = {Large Scale Kernel Machines},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	129 year = {2007},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	130 publisher = {MIT Press},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	131 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/bengio+lecun_chapter2007.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	132 abstract = {One long-term goal of machine learning research is to produce methods that
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	133 are applicable to highly complex tasks, such as perception (vision, audition), reasoning,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	134 intelligent control, and other artificially intelligent behaviors. We argue
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	135 that in order to progress toward this goal, the Machine Learning community must
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	136 endeavor to discover algorithms that can learn highly complex functions, with minimal
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	137 need for prior knowledge, and with minimal human intervention. We present
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	138 mathematical and empirical evidence suggesting that many popular approaches
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	139 to non-parametric learning, particularly kernel methods, are fundamentally limited
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	140 in their ability to learn complex high-dimensional functions. Our analysis
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	141 focuses on two problems. First, kernel machines are shallow architectures, in
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	142 which one large layer of simple template matchers is followed by a single layer
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	143 of trainable coefficients. We argue that shallow architectures can be very inefficient
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	144 in terms of required number of computational elements and examples. Second,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	145 we analyze a limitation of kernel machines with a local kernel, linked to the
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	146 curse of dimensionality, that applies to supervised, unsupervised (manifold learning)
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	147 and semi-supervised kernel machines. Using empirical results on invariant
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	148 image recognition tasks, kernel methods are compared with deep architectures, in
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	149 which lower-level features or concepts are progressively combined into more abstract
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	150 and higher-level representations. We argue that deep architectures have the
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	151 potential to generalize in non-local ways, i.e., beyond immediate neighbors, and
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	152 that this is crucial in order to make progress on the kind of complex tasks required
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	153 for artificial intelligence.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	154 cat={B},topics={HighDimensional},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	155 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	156
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	157 @ARTICLE{Bengio+Delalleau-2009,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	158 author = {Bengio, Yoshua and Delalleau, Olivier},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	159 title = {Justifying and Generalizing Contrastive Divergence},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	160 journal = {Neural Computation},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	161 volume = {21},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	162 number = {6},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	163 year = {2009},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	164 pages = {1601--1621},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	165 abstract = {We study an expansion of the log-likelihood in undirected graphical models such as the Restricted {Boltzmann} Machine (RBM), where each term in the expansion is associated with a sample in a Gibbs chain alternating between two random variables (the visible vector and the hidden vector, in RBMs). We are particularly interested in estimators of the gradient of the log-likelihood obtained through this expansion. We show that its residual term converges to zero, justifying the use of a truncation, i.e. running only a short Gibbs chain, which is the main idea behind the Contrastive Divergence (CD) estimator of the log-likelihood gradient. By truncating even more, we obtain a stochastic reconstruction error, related through a mean-field approximation to the reconstruction error often used to train autoassociators and stacked auto-associators. The derivation is not specific to the particular parametric forms used in RBMs, and only requires convergence of the Gibbs chain. We present theoretical and empirical evidence linking the number of Gibbs steps $k$ and the magnitude of the RBM parameters to the bias in the CD estimator. These experiments also suggest that the sign of the CD estimator is correct most of the time, even when the bias is large, so that CD-$k$ is a good descent direction even for small $k$.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	166 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	167
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	168 @TECHREPORT{Bengio+Delalleau-TR2007,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	169 author = {Bengio, Yoshua and Delalleau, Olivier},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	170 keywords = {Contrastive Divergence, Restricted {Boltzmann} Machine},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	171 title = {Justifying and Generalizing Contrastive Divergence},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	172 number = {1311},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	173 year = {2007},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	174 institution = {D{\'{e}}partement d'Informatique et Recherche Op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	175 abstract = {We study an expansion of the log-likelihood in undirected graphical models such as the Restricted {Boltzmann} Machine (RBM), where each term in the expansion is associated with a sample in a Gibbs chain alternating between two random variables (the visible vector and the hidden vector, in RBMs). We are particularly interested in estimators of the gradient of the log-likelihood obtained through this expansion. We show that its terms converge to zero, justifying the use of a truncation, i.e. running only a short Gibbs chain, which is the main idea behind the Contrastive Divergence approximation of the log-likelihood gradient. By truncating even more, we obtain a stochastic reconstruction error, related through a mean-field approximation to the reconstruction error often used to train autoassociators and stacked auto-associators. The derivation is not specific to the particular parametric forms used in RBMs, and only requires convergence of the Gibbs chain.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	176 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	177
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	178 @INPROCEEDINGS{Bengio+DeMori88,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	179 author = {Bengio, Yoshua and De Mori, Renato},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	180 title = {Use of neural networks for the recognition of place of articulation},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	181 booktitle = {International Conference on Acoustics, Speech and Signal Processing},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	182 year = {1988},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	183 pages = {103--106},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	184 topics={Speech},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	185 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	186
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	187 @INPROCEEDINGS{Bengio+DeMori89,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	188 author = {Bengio, Yoshua and Cardin, Regis and Cosi, Piero and De Mori, Renato},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	189 title = {Speech coding with multi-layer networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	190 booktitle = {International Conference on Acoustics, Speech and Signal Processing},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	191 year = {1989},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	192 pages = {164--167},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	193 topics={Speech},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	194 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	195
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	196 @INCOLLECTION{Bengio+DeMori90a,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	197 author = {Bengio, Yoshua and De Mori, Renato},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	198 editor = {Sethi, I. K. and Jain, A. K.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	199 title = {Connectionist models and their application to automatic speech recognition},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	200 booktitle = {Artificial Neural Networks and Statistical Pattern Recognition: Old and New Connections},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	201 year = {1990},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	202 pages = {175--192},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	203 publisher = {Elsevier, Machine Intelligence and Pattern Recognition Series},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	204 topics={Speech},cat={B},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	205 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	206
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	207 @ARTICLE{Bengio+Frasconi-jair95,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	208 author = {Bengio, Yoshua and Frasconi, Paolo},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	209 title = {Diffusion of Context and Credit Information in {M}arkovian Models},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	210 journal = {Journal of Artificial Intelligence Research},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	211 volume = {3},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	212 year = {1995},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	213 pages = {249--270},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	214 abstract = {This paper studies the problem of ergodicity of transition probability matrices in {Markovian} models, such as hidden {Markov} models ({HMM}s), and how it makes very difficult the task of learning to represent long-term context for sequential data. This phenomenon hurts the forward propagation of long-term context information, as well as learning a hidden state representation to represent long-term context, which depends on propagating credit information backwards in time. Using results from {Markov} chain theory, we show that this problem of diffusion of context and credit is reduced when the transition probabilities approach 0 or 1, i.e., the transition probability matrices are sparse and the model essentially deterministic. The results found in this paper apply to learning approaches based on continuous optimization, such as gradient descent and the Baum-Welch algorithm.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	215 topics={Markov,LongTerm},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	216 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	217
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	218 @INPROCEEDINGS{Bengio+Frasconi-nips7-diffuse,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	219 author = {Bengio, Yoshua and Frasconi, Paolo},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	220 title = {Diffusion of Credit in {M}arkovian Models},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	221 year = {1995},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	222 pages = {553--560},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	223 crossref = {NIPS7-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	224 abstract = {This paper studies the problem of diffusion in {Markovian} models, such as hidden {Markov} models ({HMM}s) and how it makes very difficult the task of learning of long-term dependencies in sequences. Using results from {Markov} chain theory, we show that the problem of diffusion is reduced if the transition probabilities approach 0 or 1. Under this condition, standard {HMM}s have very limited modeling capabilities, but input/output {HMM}s can still perform interesting computations.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	225 topics={Markov},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	226 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	227
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	228 @INPROCEEDINGS{Bengio+Frasconi-nips7-iohmms,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	229 author = {Bengio, Yoshua and Frasconi, Paolo},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	230 title = {An Input/Output {HMM} Architecture},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	231 year = {1995},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	232 pages = {427--434},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	233 crossref = {NIPS7-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	234 abstract = {We introduce a recurrent architecture having a modular structure and we formulate a training procedure based on the {EM} algorithm. The resulting model has similarities to hidden {Markov} models, but supports recurrent networks processing style and allows to exploit the supervised learning paradigm while using maximum likelihood estimation.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	235 topics={Markov},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	236 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	237
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	238 @INPROCEEDINGS{Bengio+Frasconi-nips94,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	239 author = {Bengio, Yoshua and Frasconi, Paolo},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	240 title = {Credit Assignment through Time: Alternatives to Backpropagation},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	241 year = {1994},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	242 pages = {75--82},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	243 crossref = {NIPS6-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	244 abstract = {Learning to recognize or predict sequences using long-term context has many applications. However, practical and theoretical problems are found in training recurrent neural networks to perform tasks in which input/output dependencies span long intervals. Starting from a mathematical analysis of the problem, we consider and compare alternative algorithms and architectures on tasks for which the span of the input/output dependencies can be controlled. Results on the new algorithms show performance qualitatively superior to that obtained with backpropagation.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	245 topics={LongTerm},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	246 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	247
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	248 @ARTICLE{Bengio+Pouliot90,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	249 author = {Bengio, Yoshua and Pouliot, Yannick},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	250 title = {Efficient recognition of immunoglobulin domains from amino-acid sequences using a neural network},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	251 journal = {Computer Applications in the Biosciences},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	252 volume = {6},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	253 number = {2},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	254 year = {1990},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	255 pages = {319--324},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	256 topics={Bioinformatic,PriorKnowledge},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	257 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	258
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	259 @INPROCEEDINGS{Bengio+Senecal-2003,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	260 author = {Bengio, Yoshua and S{\'{e}}n{\'{e}}cal, Jean-S{\'{e}}bastien},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	261 title = {Quick Training of Probabilistic Neural Nets by Importance Sampling},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	262 booktitle = {Proceedings of the conference on Artificial Intelligence and Statistics (AISTATS)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	263 year = {2003},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	264 abstract = {Our previous work on statistical language modeling introduced the use of probabilistic feedforward neural networks to help dealing with the curse of dimensionality. Training this model by maximum likelihood however requires for each example to perform as many network passes as there are words in the vocabulary. Inspired by the contrastive divergence model, we propose and evaluate sampling-based methods which require network passes only for the observed "positive example'' and a few sampled negative example words. A very significant speed-up is obtained with an adaptive importance sampling.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	265 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	266
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	267 @ARTICLE{Bengio+Senecal-2008,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	268 author = {Bengio, Yoshua and S{\'{e}}n{\'{e}}cal, Jean-S{\'{e}}bastien},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	269 keywords = {Energy-based models, fast training, importance sampling, language modeling, Monte Carlo methods, probabilistic neural networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	270 title = {Adaptive Importance Sampling to Accelerate Training of a Neural Probabilistic Language Model},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	271 journal = {IEEE Transactions on Neural Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	272 volume = {19},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	273 number = {4},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	274 year = {2008},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	275 pages = {713--722},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	276 abstract = {Previous work on statistical language modeling has shown that it is possible to train a feedforward neural network to approximate probabilities over sequences of words, resulting in significant error reduction when compared to standard baseline models based on -grams. However, training the neural network model with the maximum-likelihood criterion requires computations proportional to the number of words in the vocabulary. In this paper, we introduce adaptive importance sampling as a way to accelerate training of the model. The idea is to use an adaptive n-gram model to track the conditional distributions produced by the neural network. We show that a very significant speedup can be obtained on standard problems.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	277 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	278
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	279 @INCOLLECTION{Bengio-2007,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	280 author = {Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	281 editor = {Cisek, Paul and Kalaska, John and Drew, Trevor},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	282 title = {On the Challenge of Learning Complex Functions},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	283 booktitle = {Computational Neuroscience: Theoretical Insights into Brain Function},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	284 series = {Progress in Brain Research},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	285 year = {2007},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	286 publisher = {Elsevier},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	287 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/PBR_chapter.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	288 abstract = {A common goal of computational neuroscience and of artificial intelligence
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	289 research based on statistical learning algorithms is the discovery and
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	290 understanding of computational principles that could explain what we
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	291 consider adaptive intelligence, in animals as well as in machines. This
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	292 chapter focuses on what is required for the learning of complex behaviors. We
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	293 believe it involves the learning of highly varying functions, in a
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	294 mathematical sense. We bring forward two types of arguments which convey
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	295 the message that many currently popular machine learning approaches to
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	296 learning flexible functions have fundamental limitations that render them
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	297 inappropriate for learning highly varying functions. The first issue
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	298 concerns the representation of such functions with what we call shallow model
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	299 architectures. We discuss limitations of shallow architectures, such as
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	300 so-called kernel machines, boosting algorithms, and one-hidden-layer artificial neural
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	301 networks. The second issue is more focused and concerns kernel machines
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	302 with a local kernel (the type used most often in practice),
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	303 that act like a collection of template matching units. We present
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	304 mathematical results on such computational architectures showing that they
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	305 have a limitation similar to those already proved for older non-parametric
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	306 methods, and connected to the so-called curse of dimensionality. Though it has long
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	307 been believed that efficient learning in deep architectures is difficult,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	308 recently proposed computational principles for learning in deep architectures
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	309 may offer a breakthrough.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	310 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	311
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	312 @ARTICLE{Bengio-2009,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	313 author = {Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	314 title = {Learning deep architectures for {AI}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	315 journal = {Foundations and Trends in Machine Learning},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	316 volume = {2},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	317 number = {1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	318 year = {2009},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	319 pages = {1--127},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	320 note = {Also published as a book. Now Publishers, 2009.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	321 abstract = {Theoretical results suggest that in order to learn the kind of
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	322 complicated functions that can represent high-level abstractions (e.g. in
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	323 vision, language, and other AI-level tasks), one may need {\insist deep
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	324 architectures}. Deep architectures are composed of multiple levels of non-linear
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	325 operations, such as in neural nets with many hidden layers or in complicated
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	326 propositional formulae re-using many sub-formulae. Searching the
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	327 parameter space of deep architectures is a difficult task, but
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	328 learning algorithms such as those for Deep Belief Networks have recently been proposed
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	329 to tackle this problem with notable success, beating the state-of-the-art
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	330 in certain areas. This paper discusses the motivations and principles regarding
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	331 learning algorithms for deep architectures, in particular those exploiting as
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	332 building blocks unsupervised learning of single-layer models such as Restricted {Boltzmann} Machines,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	333 used to construct deeper models such as Deep Belief Networks.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	334 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	335
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	336 @TECHREPORT{Bengio-96-TR,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	337 author = {Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	338 title = {Using a Financial Training Criterion Rather than a Prediction Criterion},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	339 number = {\#1019},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	340 year = {1996},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	341 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	342 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/bengioy_TR1019.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	343 abstract = {The application of this work is to decision taking with financial time-series, using learning algorithms. The traditional approach is to train a model using a rediction criterion, such as minimizing the squared error between predictions and actual values of a dependent variable, or maximizing the likelihood of a conditional model of the dependent variable. We find here with noisy time-series that better results can be obtained when the model is directly trained in order to optimize the financial criterion of interest. Experiments were performed on portfolio selection with 35 Canadian stocks.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	344 topics={Finance,Discriminant},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	345 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	346
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	347 @BOOK{bengio-book96,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	348 author = {Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	349 title = {Neural Networks for Speech and Sequence Recognition},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	350 year = {1996},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	351 publisher = {International Thompson Computer Press},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	352 topics={Speech},cat={B},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	353 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	354
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	355 @TECHREPORT{Bengio-convex-05,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	356 author = {Bengio, Yoshua and Le Roux, Nicolas and Vincent, Pascal and Delalleau, Olivier and Marcotte, Patrice},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	357 title = {Convex neural networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	358 number = {1263},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	359 year = {2005},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	360 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	361 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/TR1263.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	362 abstract = {Convexity has recently received a lot of attention in the machine learning community, and the lack of convexity has been seen as a major disadvantage of many learning algorithms, such as multi-layer artificial neural networks. We how that training multi-layer neural networks in which the number of hidden units is learned can be viewed as a convex optimization problem. This problem involves an infinite number of variables, but can be solved by incrementally inserting a hidden unit at a time, each time finding a linear classifiers that minimizes a weighted sum of errors.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	363 topics={Boosting},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	364 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	365
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	366 @ARTICLE{Bengio-decision-trees10,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	367 author = {Bengio, Yoshua and Delalleau, Olivier and Simard, Clarence},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	368 title = {Decision Trees do not Generalize to New Variations},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	369 journal = {Computational Intelligence},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	370 year = {2010},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	371 note = {To appear}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	372 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	373
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	374 @ARTICLE{bengio-demori89,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	375 author = {Bengio, Yoshua and De Mori, Renato},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	376 title = {Use of multilayer networks for the recognition of phonetic features and phonemes},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	377 journal = {Computational Intelligence},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	378 volume = {5},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	379 year = {1989},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	380 pages = {134--141},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	381 topics={Speech},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	382 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	383
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	384 @ARTICLE{Bengio-eigen-NC2004,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	385 author = {Bengio, Yoshua and Delalleau, Olivier and Le Roux, Nicolas and Paiement, Jean-Fran{\c c}ois and Vincent, Pascal and Ouimet, Marie},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	386 title = {Learning eigenfunctions links spectral embedding and kernel {PCA}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	387 journal = {Neural Computation},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	388 volume = {16},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	389 number = {10},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	390 year = {2004},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	391 pages = {2197--2219},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	392 abstract = {In this paper, we show a direct relation between spectral embedding methods and kernel {PCA}, and how both are special cases of a more general learning problem, that of learning the principal eigenfunctions of an operator defined from a kernel and the unknown data generating density. Whereas spectral embedding methods only provided coordinates for the training points, the analysis justifies a simple extension to out-of-sample examples (the Nystr{\"{o}}m formula) for Multi-Dimensional Scaling, spectral clustering, Laplacian eigenmaps, Locally Linear Embedding ({LLE}) and Isomap. The analysis provides, for all such spectral embedding methods, the definition of a loss function, whose empirical average is minimized by the traditional algorithms. The asymptotic expected value of that loss defines a generalization performance and clarifies what these algorithms are trying to learn. Experiments with {LLE}, Isomap, spectral clustering and {MDS} show that this out-of-sample embedding formula generalizes well, with a level of error comparable to the effect of small perturbations of the training set on the embedding.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	393 topics={HighDimensional,Kernel,Unsupervised},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	394 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	395
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	396 @INPROCEEDINGS{Bengio-Gingras-nips8,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	397 author = {Bengio, Yoshua and Gingras, Fran{\c c}ois},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	398 title = {Recurrent Neural Networks for Missing or Asynchronous Data},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	399 year = {1996},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	400 pages = {395--401},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	401 crossref = {NIPS8-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	402 abstract = {In this paper we propose recurrent neural networks with feedback into the input units for handling two types of data analysis problems. On the one hand, this scheme can be used for static data when some of the input variables are missing. On the other hand, it can also be used for sequential data, when some of the input variables are missing or are available at different frequencies. Unlike in the case of probabilistic models (e.g. Gaussian) of the missing variables, the network does not attempt to model the distribution of the missing variables given the observed variables. Instead it is a more discriminant approach that fills in the missing variables for the sole purpose of minimizing a learning criterion (e.g., to minimize an output error).},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	403 topics={Finance,Missing},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	404 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	405
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	406 @ARTICLE{Bengio-Grandvalet-JMLR-04,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	407 author = {Bengio, Yoshua and Grandvalet, Yves},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	408 title = {No Unbiased Estimator of the Variance of K-Fold Cross-Validation},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	409 volume = {5},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	410 year = {2004},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	411 pages = {1089--1105},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	412 journal = {Journal of Machine Learning Research},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	413 abstract = {Most machine learning researchers perform quantitative experiments to estimate generalization error and compare the performance of different algorithms (in particular, their proposed algorithm). In order to be able to draw statistically convincing conclusions, it is important to estimate the uncertainty of such estimates. This paper studies the very commonly used K-fold cross-validation estimator of generalization performance. The main theorem shows that there exists no universal (valid under all distributions) unbiased estimator of the variance of K-fold cross-validation. The analysis that accompanies this result is based on the eigen-decomposition of the covariance matrix of errors, which has only three different eigenvalues corresponding to three degrees of freedom of the matrix and three components of the total variance. This analysis helps to better understand the nature of the problem and how it can make naive estimators (that dont take into account the error correlations due to the overlap between training and test sets) grossly underestimate variance. This is confirmed by numerical experiments in which the three components of the variance are compared when the difficulty of the learning problem and the number of folds are varied.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	414 topics={Comparative},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	415 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	416
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	417 @TECHREPORT{bengio-hyper-TR99,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	418 author = {Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	419 title = {Continuous Optimization of Hyper-Parameters},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	420 number = {1144},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	421 year = {1999},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	422 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	423 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/hyperTR.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	424 abstract = {Many machine learning algorithms can be formulated as the minimization of a training criterion which involves (1) training errors on each training example and (2) some hyper-parameters, which are kept fixed during this minimization. When there is only a single hyper-parameter one can easily explore how its value aects a model selection criterion (that is not the same as the training criterion, and is used to select hyper-parameters). In this paper we present a methodology to select many hyper-parameters that is based on the computation of the gradient of a model selection criterion with respect to the hyper-parameters. We first consider the case of a training criterion that is quadratic in the parameters. In that case, the gradient of the selection criterion with respect to the hyper-parameters is efficiently computed by back-propagating through a Cholesky decomposition. In the more general case, we show that the implicit function theorem can be used to derive a formula for the hyper-parameter gradient, but this formula requires the computation of second derivatives of the training criterion},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	425 topics={ModelSelection},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	426 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	427
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	428 @INPROCEEDINGS{Bengio-icnn93,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	429 author = {Bengio, Yoshua and Frasconi, Paolo and Simard, Patrice},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	430 title = {The problem of learning long-term dependencies in recurrent networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	431 booktitle = {IEEE International Conference on Neural Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	432 year = {1993},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	433 pages = {1183--1195},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	434 publisher = {IEEE Press},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	435 note = {(invited paper)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	436 topics={LongTerm},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	437 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	438
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	439 @ARTICLE{Bengio-ijprai93,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	440 author = {Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	441 title = {A Connectionist Approach to Speech Recognition},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	442 journal = {International Journal on Pattern Recognition and Artificial Intelligence},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	443 volume = {7},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	444 number = {4},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	445 year = {1993},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	446 pages = {647--668},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	447 abstract = {The task discussed in this paper is that of learning to map input sequences to output sequences. In particular, problems of phoneme recognition in continuous speech are considered, but most of the discussed techniques could be applied to other tasks, such as the recognition of sequences of handwritten characters. The systems considered in this paper are based on connectionist models, or artificial neural networks, sometimes combined with statistical techniques for recognition of sequences of patterns, stressing the integration of prior knowledge and learning. Different architectures for sequence and speech recognition are reviewed, including recurrent networks as well as hybrid systems involving hidden {Markov} models.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	448 topics={PriorKnowledge,Speech},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	449 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	450
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	451 @TECHREPORT{Bengio-iohmms-TR99,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	452 author = {Bengio, Yoshua and Lauzon, Vincent-Philippe and Ducharme, R{\'{e}}jean},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	453 title = {Experiments on the Application of {IOHMM}s to Model Financial Returns Series},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	454 number = {1146},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	455 year = {1999},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	456 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	457 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/iohmms-returnsTR.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	458 abstract = {Input/Output Hidden {Markov} Models ({IOHMM}s) are conditional hidden {Markov} models in which the emission (and possibly the transition) probabilities can be conditionned on an input sequence. For example, these conditional distributions can be linear, logistic, or non-linear (using for example multi-layer neural networks). We compare the generalization performance of several models which are special cases of Input/Output Hidden {Markov} Models on financial time-series prediction tasks: an unconditional Gaussian, a conditional linear Gaussian, a mixture of Gaussians, a mixture of conditional linear Gaussians, a hidden {Markov} model, and various {IOHMM}s. The experiments are performed on modeling the returns of market and sector indices. Note that the unconditional Gaussian estimates the first moment with the historical average. The results show that, although for the first moment the historical average gives the best results, for the higher moments, the {IOHMM}s yielded significantly better performance, as measured by the out-of-sample likelihood.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	459 topics={Markov},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	460 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	461
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	462 @ARTICLE{bengio-lauzon-ducharme:2000,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	463 author = {Bengio, Yoshua and Lauzon, Vincent-Philippe and Ducharme, R{\'{e}}jean},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	464 title = {Experiments on the Application of {IOHMM}s to Model Financial Returns Series},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	465 journal = {IEEE Transaction on Neural Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	466 volume = {12},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	467 number = {1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	468 year = {2001},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	469 pages = {113--123},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	470 abstract = {Input/Output Hidden {Markov} Models ({IOHMM}s) are conditional hidden {Markov} models in which the emission (and possibly the transition) probabilities can be conditioned on an input sequence. For example, these conditional distributions can be logistic, or non-linear (using for example multi-layer neural networks). We compare generalization performance of several models which are special cases of Input/Output Hidden {Markov} Models on financial time-series prediction tasks: an unconditional Gaussian, a conditional linear Gaussian, a mixture of Gaussians, a mixture of conditional linear Gaussians, a hidden {Markov} model, and various {IOHMM}s. The experiments compare these models on predicting the conditional density of returns of market sector indices. Note that the unconditional Gaussian estimates the first moment the historical average. The results show that_ although for the first moment the historical average gives the best results, for the higher moments, the {IOHMM}s significantly better performance, as estimated by the out-of-sample likelihood.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	471 topics={Markov,Finance},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	472 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	473
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	474 @INPROCEEDINGS{bengio-lecun-94,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	475 author = {Bengio, Yoshua and {LeCun}, Yann},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	476 title = {Word normalization for on-line handwritten word recognition},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	477 booktitle = {Proc. of the International Conference on Pattern Recognition},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	478 volume = {II},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	479 year = {1994},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	480 pages = {409--413},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	481 publisher = {IEEE},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	482 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/icpr-norm.ps},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	483 abstract = {We introduce a new approach to normalizing words written with an electronic stylus that applies to all styles of handwriting (upper case, lower case, printed, cursive, or mixed). A geometrical model of the word spatial structure is fitted to the pen trajectory using the {EM} algorithm. The fitting process maximizes the likelihood of the trajectory given the model and a set a priors on its parameters. The method was evaluated and integrated to a recognition system that combines neural networks and hidden {Markov} models.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	484 topics={PriorKnowledge,Speech},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	485 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	486
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	487 @TECHREPORT{Bengio-localfailure-TR-2005,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	488 author = {Bengio, Yoshua and Delalleau, Olivier and Le Roux, Nicolas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	489 title = {The Curse of Dimensionality for Local Kernel Machines},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	490 number = {1258},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	491 year = {2005},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	492 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	493 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/tr1258.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	494 abstract = {We present a series of theoretical arguments supporting the claim that a large class of modern learning algorithms based on local kernels are sensitive to the curse of dimensionality. These include local manifold learning algorithms such as Isomap and {LLE}, support vector classifiers with Gaussian or other local kernels, and graph-based semisupervised learning algorithms using a local similarity function. These algorithms are shown to be local in the sense that crucial properties of the learned function at x depend mostly on the neighbors of x in the training set. This makes them sensitive to the curse of dimensionality, well studied for classical non-parametric statistical learning. There
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	495 is a large class of data distributions for which non-local solutions could be expressed compactly and potentially be learned with few examples, but which will require a large number of local bases and therefore a large number of training examples when using a local learning algorithm.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	496 topics={HighDimensional,Kernel,Unsupervised},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	497 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	498
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	499 @INPROCEEDINGS{Bengio-nips-2006,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	500 author = {Bengio, Yoshua and Lamblin, Pascal and Popovici, Dan and Larochelle, Hugo},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	501 title = {Greedy Layer-Wise Training of Deep Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	502 year = {2007},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	503 pages = {153--160},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	504 crossref = {NIPS19-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	505 abstract = {Complexity theory of circuits strongly suggests that deep architectures can be
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	506 much more efficient (sometimes exponentially) than shallow architectures,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	507 in terms of computational elements required to represent some functions.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	508 Deep multi-layer neural networks have many levels of non-linearities
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	509 allowing them to compactly represent highly non-linear and
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	510 highly-varying functions. However, until recently it was not clear how
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	511 to train such deep networks, since gradient-based
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	512 optimization starting from random initialization appears to often get stuck
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	513 in poor solutions. Hinton et al. recently introduced
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	514 a greedy layer-wise unsupervised learning algorithm for Deep Belief
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	515 Networks (DBN), a generative model with many layers of hidden causal
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	516 variables. In the context of the above optimization problem,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	517 we study this algorithm empirically and explore variants to
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	518 better understand its success and extend it to cases where the inputs are
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	519 continuous or where the structure of the input distribution is not
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	520 revealing enough about the variable to be predicted in a supervised task.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	521 Our experiments also confirm the hypothesis that the greedy
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	522 layer-wise unsupervised training strategy mostly helps the
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	523 optimization, by initializing weights in a region near a
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	524 good local minimum, giving rise to internal distributed representations
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	525 that are high-level abstractions of the input, bringing better generalization.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	526 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	527
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	528 @INPROCEEDINGS{Bengio-nips10,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	529 author = {Bengio, Yoshua and Bengio, Samy and Isabelle, Jean-Fran{\c c}ois and Singer, Yoram},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	530 title = {Shared Context Probabilistic Transducers},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	531 year = {1998},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	532 crossref = {NIPS10-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	533 abstract = {Recently, a model for supervised learning of probabilistic transducers represented by suffix trees was introduced. However, this algorithm tends to build very large trees, requiring very large amounts of computer memory. In this paper, we propose a new, more compact, transducer model in which one shares the parameters of distributions associated to contexts yielding similar conditional output distributions. We illustrate the advantages of the proposed algorithm with comparative experiments on inducing a noun phrase recognizer.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	534 topics={HighDimensional},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	535 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	536
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	537 @TECHREPORT{Bengio-NLMP-TR-2005,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	538 author = {Bengio, Yoshua and Larochelle, Hugo},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	539 title = {Non-Local Manifold Parzen Windows},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	540 number = {1264},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	541 year = {2005},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	542 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	543 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/NLMP-techreport.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	544 abstract = {In order to escape from the curse of dimensionality, we claim that one can learn non-local functions, in the sense that the value and shape of the learned function at x must be inferred using examples that may be far from x. With this objective, we present a non-local non-parametric density estimator. It builds upon previously proposed Gaussian mixture models with regularized covariance matrices to take into account the local shape of the manifold. It also builds upon recent work on non-local estimators of the tangent plane of a manifold, which are able to generalize in places with little training data, unlike traditional, local, non-parametric models.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	545 topics={HighDimensional,Kernel,Unsupervised},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	546 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	547
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	548 @INPROCEEDINGS{Bengio-nncm96,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	549 author = {Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	550 editor = {Weigend, A.S. and Abu-Mostafa, Y.S. and Refenes, A. -P. N.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	551 title = {Training A Neural Network with a Financial Criterion Rather than a Prediction Criterion},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	552 booktitle = {Proceedings of the Fourth International Conference on Neural Networks in the Capital Markets ({NNCM}-96)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	553 year = {1997},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	554 pages = {433--443},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	555 publisher = {World Scientific},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	556 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/nncm.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	557 abstract = {A common approach to quantitative decision taking with financial time-series is to train a model using a prediction criterion (e.g., squared error). We find on a portfolio selection problem that better results can be obtained when the model is directly trained in order to optimize the financial criterion of interest, with a differentiable decision module.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	558 topics={Finance,PriorKnowledge,Discriminant},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	559 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	560
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	561 @TECHREPORT{Bengio-NonStat-Hyper-TR,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	562 author = {Bengio, Yoshua and Dugas, Charles},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	563 title = {Learning Simple Non-Stationarities with Hyper-Parameters},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	564 number = {1145},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	565 year = {1999},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	566 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	567 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/nonstatTR.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	568 abstract = {We consider sequential data that is sampled from an unknown process, so that the data are not necessarily i.i.d.. Most approaches to machine learning assume that data points are i.i.d.. Instead we consider a measure of generalization that does not make this assumption, and we consider in this context a recently proposed approach to optimizing hyper-parameters, based on the computation of the gradient of a model selection criterion with respect to hyper-parameters. Here we use hyper-parameters that control a function that gives different weights to different time steps in the historical data sequence. The approach is successfully applied to modeling thev olatility of stock returns one month ahead. Comparative experiments with more traditional methods are presented.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	569 topics={ModelSelection,Finance},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	570 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	571
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	572 @ARTICLE{Bengio-scholarpedia-2007,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	573 author = {Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	574 title = {Neural net language models},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	575 journal = {Scholarpedia},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	576 volume = {3},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	577 number = {1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	578 year = {2008},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	579 pages = {3881},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	580 abstract = {A language model is a function, or an algorithm for learning such a function, that captures the salient statistical characteristics of the distribution of sequences of words in a natural language, typically allowing one to make probabilistic predictions of the next word given preceding ones.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	581
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	582 A neural network language model is a language model based on Neural Networks , exploiting their ability to learn distributed representations to reduce the impact of the curse of dimensionality.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	583
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	584 In the context of learning algorithms, the curse of dimensionality refers to the need for huge numbers of training examples when learning highly complex functions. When the number of input variables increases, the number of required examples can grow exponentially. The curse of dimensionality arises when a huge number of different combinations of values of the input variables must be discriminated from each other, and the learning algorithm needs at least one example per relevant combination of values. In the context of language models, the problem comes from the huge number of possible sequences of words, e.g., with a sequence of 10 words taken from a vocabulary of 100,000 there are 10^{50} possible sequences...
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	585
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	586 A distributed representation of a symbol is a tuple (or vector) of features which characterize the meaning of the symbol, and are not mutually exclusive. If a human were to choose the features of a word, he might pick grammatical features like gender or plurality, as well as semantic features like animate" or invisible. With a neural network language model, one relies on the learning algorithm to discover these features, and the features are continuous-valued (making the optimization problem involved in learning much simpler).
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	587
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	588 The basic idea is to learn to associate each word in the dictionary with a continuous-valued vector representation. Each word corresponds to a point in a feature space. One can imagine that each dimension of that space corresponds to a semantic or grammatical characteristic of words. The hope is that functionally similar words get to be closer to each other in that space, at least along some directions. A sequence of words can thus be transformed into a sequence of these learned feature vectors. The neural network learns to map that sequence of feature vectors to a prediction of interest, such as the probability distribution over the next word in the sequence. What pushes the learned word features to correspond to a form of semantic and grammatical similarity is that when two words are functionally similar, they can be replaced by one another in the same context, helping the neural network to compactly represent a function that makes good predictions on the training set, the set of word sequences used to train the model.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	589
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	590 The advantage of this distributed representation approach is that it allows the model to generalize well to sequences that are not in the set of training word sequences, but that are similar in terms of their features, i.e., their distributed representation. Because neural networks tend to map nearby inputs to nearby outputs, the predictions corresponding to word sequences with similar features are mapped to similar predictions. Because many different combinations of feature values are possible, a very large set of possible meanings can be represented compactly, allowing a model with a comparatively small number of parameters to fit a large training set.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	591 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	592
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	593 @TECHREPORT{Bengio-TR1312,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	594 author = {Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	595 title = {Learning deep architectures for AI},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	596 number = {1312},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	597 year = {2007},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	598 institution = {Dept. IRO, Universite de Montreal},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	599 note = {Preliminary version of journal article with the same title appearing in Foundations and Trends in Machine Learning (2009)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	600 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/TR1312.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	601 abstract = {Theoretical results strongly suggest that in order to learn the kind of
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	602 complicated functions that can represent high-level abstractions (e.g. in
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	603 vision, language, and other AI-level tasks), one may need deep
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	604 architectures. Deep architectures are composed of multiple levels of non-linear
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	605 operations, such as in neural nets with many hidden layers. Searching the
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	606 parameter space of deep architectures is a difficult optimization task, but
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	607 learning algorithms such as those for Deep Belief Networks have recently been proposed
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	608 to tackle this problem with notable success, beating the state-of-the-art
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	609 in certain areas. This paper discusses the motivations and principles regarding
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	610 learning algorithms for deep architectures and in particular for those based
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	611 on unsupervised learning such as Deep Belief Networks, using as building
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	612 blocks single-layer models such as Restricted {Boltzmann} Machines.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	613 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	614
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	615 @ARTICLE{Bengio-trnn94,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	616 author = {Bengio, Yoshua and Simard, Patrice and Frasconi, Paolo},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	617 title = {Learning Long-Term Dependencies with Gradient Descent is Difficult},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	618 journal = {IEEE Transactions on Neural Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	619 volume = {5},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	620 number = {2},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	621 year = {1994},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	622 pages = {157--166},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	623 abstract = {Recurrent neural networks can be used to map input sequences to output sequences, such as for recognition, production or prediction problems. However, practical difficulties have been reported in training recurrent neural networks to perform tasks in which the temporal contingencies present in the input/output sequences span long intervals. We show why gradient based learning algorithms face an increasingly difficult problem as the duration of the dependencies to be captures increases. These results expose a trade-off between efficient learning by gradient descent and latching on information for long periods. Based on an understanding of this problem, alternatives to standard gradient descent are considered.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	624 optnote={(Special Issue on Recurrent Neural Networks)},topics={LongTerm},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	625 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	626
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	627 @INPROCEEDINGS{Bengio-wirn93,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	628 author = {Bengio, Yoshua and Frasconi, Paolo and Gori, Marco and Soda, G.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	629 editor = {Caianello, E.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	630 title = {Recurrent Neural Networks for Adaptive Temporal Processing},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	631 booktitle = {Proc. of the 6th Italian Workshop on Neural Networks, WIRN-93},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	632 year = {1993},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	633 pages = {1183--1195},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	634 publisher = {World Scientific Publ.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	635 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/rnn_review93.ps},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	636 topics={LongTerm},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	637 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	638
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	639 @ARTICLE{Bengio2000c,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	640 author = {Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	641 title = {Gradient-Based Optimization of Hyperparameters},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	642 journal = {Neural Computation},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	643 volume = {12},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	644 number = {8},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	645 year = {2000},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	646 pages = {1889--1900},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	647 abstract = {Many machine learning algorithms can be formulated as the minimization of a training criterion which involves a hyper-parameter. This hyper-parameter is usually chosen by trial and error with a model selection criterion. In this paper we present a methodology to optimize several hyper-parameters, based on the computation of the gradient of a model selection criterion with respect to the hyper-parameters. In the case of a quadratic training criterion, the gradient of the selection criterion with respect to the hyper-parameters is efficiently computed by back-propagating through a Cholesky decomposition. In the more general case, we show that the implicit function theorem can be used to derive a formula for the hyper-parameter gradient involving second derivatives of the training criterion.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	648 topics={ModelSelection},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	649 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	650
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	651 @ARTICLE{Bengio89a,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	652 author = {Bengio, Yoshua and Cardin, Regis and De Mori, Renato and Merlo, Ettore},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	653 title = {Programmable execution of multi-layered networks for automatic speech recognition},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	654 journal = {Communications of the Association for Computing Machinery},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	655 volume = {32},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	656 number = {2},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	657 year = {1989},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	658 pages = {195--199},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	659 topics={Speech},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	660 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	661
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	662 @INPROCEEDINGS{Bengio89c,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	663 author = {Bengio, Yoshua and Cosi, Piero and Cardin, Regis and De Mori, Renato},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	664 title = {Use of multi-layered networks for coding speech with phonetic features},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	665 year = {1989},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	666 pages = {224--231},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	667 crossref = {NIPS1-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	668 abstract = {Preliminary results on speaker-independant speech recognition are reported. A method that combines expertise on neural networks with expertise on speech recognition is used to build the recognition systems. For transient sounds, event-driven property extractors with variable resolution in the time and frequency domains are used. For sonorant speech, a model of the human auditory system is preferred to FFT as a front-end module.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	669 topics={Speech},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	670 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	671
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	672 @INPROCEEDINGS{Bengio89d,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	673 author = {De Mori, Renato and Bengio, Yoshua and Cosi, Piero},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	674 title = {On the generalization capability of multilayered networks in the extraction of speech properties},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	675 booktitle = {Proceedings of the International Joint Conference on Artificial Intelligence},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	676 year = {1989},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	677 pages = {1531--1536},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	678 publisher = {IEEE},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	679 topics={Speech},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	680 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	681
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	682 @INPROCEEDINGS{Bengio90,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	683 author = {Bengio, Yoshua and Cardin, Regis and De Mori, Renato},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	684 title = {Speaker Independent Speech Recognition with Neural Networks and Speech Knowledge},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	685 year = {1990},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	686 pages = {218--225},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	687 crossref = {NIPS2-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	688 abstract = {We attempt to combine neural networks with knowledge from speech science to build a speaker independent speech recognition system. This knowledge is utilized in designing the preprocessing, input coding, output coding, output supervision and architectural constraints. To handle the temporal aspect of speech we combine delays, copies of activations of hidden and output units at the input level, and Back-Propagation for Sequences (BPS), a learning algorithm for networks with local self-loops. This strategy is demonstrated in several experiments, in particular a nasal discrimination task for which the application of a speech theory hypothesis dramatically improved generalization.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	689 topics={PriorKnowledge,Speech},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	690 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	691
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	692 @INCOLLECTION{Bengio90b,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	693 author = {Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	694 title = {Radial Basis Functions for speech recognition},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	695 booktitle = {Speech Recognition and Understanding: Recent Advances, Trends and Applications},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	696 year = {1990},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	697 pages = {293--298},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	698 publisher = {NATO Advanced Study Institute Series F: Computer and Systems Sciences},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	699 topics={Kernel,Speech},cat={B},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	700 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	701
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	702 @INCOLLECTION{Bengio90c,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	703 author = {Bengio, Yoshua and De Mori, Renato},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	704 editor = {{Fogelman Soulie}, F. and Herault, J.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	705 title = {Speech coding with multilayer networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	706 booktitle = {Neurocomputing: Algorithms, Architectures and Applications},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	707 year = {1990},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	708 pages = {207--216},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	709 publisher = {NATO Advanced Study Institute Series F: Computer and Systems Sciences},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	710 topics={Speech},cat={B},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	711 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	712
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	713 @INPROCEEDINGS{Bengio90e,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	714 author = {Bengio, Yoshua and Pouliot, Yannick and Bengio, Samy and Agin, Patrick},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	715 title = {A neural network to detect homologies in proteins},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	716 year = {1990},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	717 pages = {423--430},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	718 crossref = {NIPS2-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	719 abstract = {In order to detect the presence and location of immunoglobulin (Ig) domains from amino acid sequences we built a system based on a neural network with one hidden layer trained with back propagation. The program was designed to efficiently identify proteins exhibiting such domains, characterized by a few localized conserved regions and a low overall homology. When the National Biomedical Research Foundation (NBRF) NEW protein sequence database was scanned to evaluate the program's performance, we obtained very low rates of false negatives coupled with a moderate rate of false positives.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	720 topics={Bioinformatic,PriorKnowledge},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	721 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	722
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	723 @INPROCEEDINGS{Bengio90z,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	724 author = {Bengio, Yoshua and De Mori, Renato and Gori, Marco},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	725 editor = {Caianello, E.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	726 title = {Experiments on automatic speech recognition using BPS},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	727 booktitle = {Parallel Architectures and Neural Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	728 year = {1990},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	729 pages = {223--232},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	730 publisher = {World Scientific Publ.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	731 topics={Speech},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	732 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	733
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	734 @INPROCEEDINGS{Bengio91a,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	735 author = {Bengio, Yoshua and De Mori, Renato and Flammia, Giovanni and Kompe, Ralf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	736 title = {A comparative study of hybrid acoustic phonetic decoders based on artificial neural networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	737 booktitle = {Proceedings of EuroSpeech'91},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	738 year = {1991},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	739 topics={PriorKnowledge,Speech},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	740 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	741
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	742 @INPROCEEDINGS{Bengio91b,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	743 author = {Bengio, Yoshua and De Mori, Renato and Flammia, Giovanni and Kompe, Ralf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	744 title = {Global Optimization of a Neural Network - Hidden {M}arkov Model Hybrid},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	745 booktitle = {Proceedings of EuroSpeech'91},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	746 year = {1991},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	747 topics={Markov},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	748 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	749
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	750 @INPROCEEDINGS{Bengio91z,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	751 author = {Bengio, Yoshua and De Mori, Renato and Flammia, Giovanni and Kompe, Ralf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	752 title = {Phonetically motivated acoustic parameters for continuous speech recognition using artificial neural networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	753 booktitle = {Proceedings of EuroSpeech'91},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	754 year = {1991},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	755 cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	756 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	757
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	758 @ARTICLE{Bengio92b,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	759 author = {Bengio, Yoshua and De Mori, Renato and Flammia, Giovanni and Kompe, Ralf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	760 title = {Phonetically motivated acoustic parameters for continuous speech recognition using artificial neural networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	761 journal = {Speech Communication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	762 volume = {11},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	763 number = {2--3},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	764 year = {1992},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	765 pages = {261--271},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	766 note = {Special issue on neurospeech},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	767 topics={PriorKnowledge,Speech},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	768 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	769
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	770 @INPROCEEDINGS{Bengio92c,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	771 author = {Bengio, Yoshua and De Mori, Renato and Flammia, Giovanni and Kompe, Ralf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	772 title = {Neural Network - Gaussian Mixture Hybrid for Speech Recognition or Density Estimation},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	773 year = {1992},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	774 pages = {175--182},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	775 crossref = {NIPS4-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	776 abstract = {The subject of this paper is the integration of multi-layered Artificial Neural Networks ({ANN}) with probability density functions such as Gaussian mixtures found in continuous density hlidden {Markov} Models ({HMM}). In the first part of this paper we present an {ANN}/HMM hybrid in which all the parameters or the the system are simultaneously optimized with respect to a single criterion. In the second part of this paper, we study the relationship between the density of the inputs of the network and the density of the outputs of the networks. A rew experiments are presented to explore how to perform density estimation with {ANN}s.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	777 topics={Speech},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	778 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	779
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	780 @INPROCEEDINGS{Bengio94d,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	781 author = {Frasconi, Paolo and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	782 title = {An {EM} Approach to Grammatical Inference: Input/Output {HMMs}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	783 booktitle = {International Conference on Pattern Recognition (ICPR'94)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	784 year = {1994},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	785 pages = {289--294},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	786 topics={Markov,LongTerm},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	787 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	788
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	789 @ARTICLE{Bengio96,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	790 author = {Bengio, Yoshua and Frasconi, Paolo},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	791 title = {Input/{O}utput {HMM}s for Sequence Processing},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	792 journal = {IEEE Transactions on Neural Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	793 volume = {7},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	794 number = {5},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	795 year = {1996},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	796 pages = {1231--1249},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	797 abstract = {We consider problems of sequence processing and propose a solution based on a discrete state model in order to represent past context. We introduce a recurrent connectionist architecture having a modular structure that associates a subnetwork to each state. The model has a statistical interpretation we call Input/Output Hidden {Markov} Model ({IOHMM}). It can be trained by the {EM} or {GEM} algorithms, considering state trajectories as missing data, which decouples temporal credit assignment and actual parameter estimation.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	798 The model presents similarities to hidden {Markov} models ({HMM}s), but allows us to map input sequences to output sequences, using the same processing style as recurrent neural networks. {IOHMM}s are trained using a more discriminant learning paradigm than {HMM}s, while potentially taking advantage of the {EM} algorithm.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	799 We demonstrate that {IOHMM}s are well suited for solving grammatical inference problems on a benchmark problem. Experimental results are presented for the seven Tomita grammars, showing that these adaptive models can attain excellent generalization.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	800 topics={Markov},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	801 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	802
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	803 @TECHREPORT{Bengio96-hmmsTR,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	804 author = {Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	805 title = {Markovian Models for Sequential Data},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	806 number = {1049},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	807 year = {1996},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	808 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	809 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/hmmsTR.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	810 abstract = {Hidden {Markov} Models ({HMM}s) are statistical models of sequential data that have been used successfully in many applications, especially for speech recognition. We first summarize the basics of {HMM}s, and then review several recent related learning algorithms and extensions of {HMM}s, including hybrids of {HMM}s with artificial neural networks, Input-Output {HMM}s, weighted transducers, variable-length {Markov} models and {Markov} switching state-space models. Finally, we discuss some of the challenges of future research in this area.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	811 topics={Markov},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	812 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	813
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	814 @ARTICLE{Bengio97,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	815 author = {Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	816 title = {Using a Financial Training Criterion Rather than a Prediction Criterion},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	817 journal = {International Journal of Neural Systems},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	818 volume = {8},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	819 number = {4},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	820 year = {1997},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	821 pages = {433--443},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	822 note = {Special issue on noisy time-series},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	823 abstract = {The application of this work is to decision taking with financial time-series, using learning algorithms. The traditional approach is to train a model using a prediction criterion, such as minimizing the squared error between predictions and actual values of a dependent variable, or maximizing the likelihood of a conditional model of the dependent variable. We find here with noisy time-series that better results can be obtained when the model is directly trained in order to maximize the financial criterion of interest, here gains and losses (including those due to transactions) incurred during trading. Experiments were performed on portfolio selection with 35 Canadian stocks.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	824 topics={Finance,PriorKnowledge,Discriminant},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	825 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	826
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	827 @ARTICLE{Bengio99a,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	828 author = {Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	829 title = {Markovian Models for Sequential Data},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	830 journal = {Neural Computing Surveys},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	831 volume = {2},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	832 year = {1999},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	833 pages = {129--162},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	834 abstract = {Hidden {Markov} Models ({HMM}s) are statistical models of sequential data that have been used successfully in many machine learning applications, especially for speech recognition. Furthermore? in the last few years, many new and promising probabilistic models related to {HMM}s have been proposed. We first summarize the basics of {HMM}s, arid then review several recent related learning algorithms and extensions of {HMM}s, including in particular hybrids of {HMM}s with artificial neural networks, Input-Output {HMM}s (which are conditional {HMM}s using neural networks to compute probabilities), weighted transducers, variable-length {Markov} models and {Markov} switching state-space models. Finally, we discuss some of the challenges of future research in this very active area.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	835 topics={Markov},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	836 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	837
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	838 @ARTICLE{Bengio99b,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	839 author = {Bengio, Samy and Bengio, Yoshua and Robert, Jacques and B{\'{e}}langer, Gilles},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	840 title = {Stochastic Learning of Strategic Equilibria for Auctions},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	841 journal = {Neural Computation},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	842 volume = {11},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	843 number = {5},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	844 year = {1999},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	845 pages = {1199--1209},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	846 abstract = {This paper presents a new application of stochastic adaptive learning algorithms to the computation of strategic equilibria in auctions. The proposed approach addresses the problems of tracking a moving target and balancing exploration (of action space) versus exploitation (of better modeled regions of action space). Neural networks are used to represent a stochastic decision model for each bidder. Experiments confirm the correctness and usefulness of the approach.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	847 topics={Auction},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	848 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	849
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	850 @TECHREPORT{bengio:1990,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	851 author = {Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	852 title = {Learning a Synaptic Learning Rule},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	853 number = {751},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	854 year = {1990},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	855 institution = {D{\'{e}}partement d'Informatique et de Recherche Op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	856 topics={BioRules},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	857 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	858
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	859 @INPROCEEDINGS{bengio:1990:snowbird,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	860 author = {Bengio, Yoshua and R., De Mori},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	861 title = {Recurrent networks with Radial Basis Functions for speech recognition},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	862 booktitle = {1990 Neural Networks for Computing Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	863 year = {1990},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	864 topics={Speech},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	865 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	866
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	867 @INPROCEEDINGS{bengio:1991:ijcnn,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	868 author = {Bengio, Yoshua and Bengio, Samy and Cloutier, Jocelyn},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	869 title = {Learning a Synaptic Learning Rule},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	870 booktitle = {Proceedings of the International Joint Conference on Neural Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	871 year = {1991},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	872 pages = {II--A969},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	873 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/bengio_1991_ijcnn.ps},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	874 abstract = {This paper presents an original approach to neural modeling based on the idea of searching, with learning methods, for a synaptic learning rule which is biologically plausible, and yields networks that are able to learn to perform difficult tasks. The proposed method of automatically finding the learning rule relies on the idea of considering the synaptic modification rule as a parametric function. This function has local inputs and is the same in many neurons. The parameters that define this function can be estimated with known learning methods. For this optimization, we give particular attention to gradient descent and genetic algorithms. In both cases, estimation of this function consists of a joint global optimization of (a) the synaptic modification function, and (b) the networks that are learning to perform some tasks. The proposed methodology can be used as a tool to explore the missing pieces of the puzzle of neural networks learning. Both network architecture, and the learning function can be designed within constraints derived from biological knowledge.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	875 addressfr={Seattle, USA},topics={BioRules},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	876 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	877
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	878 @INPROCEEDINGS{bengio:1991:nnc,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	879 author = {Bengio, Yoshua and Bengio, Samy and Cloutier, Jocelyn},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	880 title = {Learning Synaptic Learning Rules},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	881 booktitle = {Neural Networks for Computing},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	882 year = {1991},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	883 addressfr={Snowbird, Utah, USA},topics={BioRules},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	884 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	885
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	886 @INPROCEEDINGS{bengio:1991:snowbird,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	887 author = {Bengio, Yoshua and Bengio, Samy and Cloutier, Jocelyn},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	888 title = {Learning a Synaptic Learning Rule},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	889 booktitle = {1991 Neural Networks for Computing Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	890 year = {1991},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	891 topics={BioRules},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	892 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	893
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	894 @INPROCEEDINGS{bengio:1992:nn,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	895 author = {Bengio, Samy and Bengio, Yoshua and Cloutier, Jocelyn and Gecsei, Jan},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	896 title = {Aspects th{\'{e}}oriques de l'optimisation d'une r{\`{e}}gle d'apprentissage},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	897 booktitle = {Actes de la conf{\'{e}}rence Neuro-N{\^{\i}}mes 1992},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	898 year = {1992},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	899 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/bengio_1992_nn.ps},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	900 abstract = {Ayant expos{\'{e}} dans de pr{\'{e}}c{\'{e}}dentes publications (voir [Beng90, Beng92] notamment) lid{\'{e}}e que lon pouvait optimiser des r{\`{e}}gles dapprentissage param{\'{e}}triques pour r{\'{e}}seaux de neurones, nous montrons dans cet article comment d{\'{e}}velopper, par la m{\'{e}}thode du Lagrangien, le gradient n{\'{e}}cessaire {\`{a}} loptimisation dune r{\`{e}}gle dapprentissage par descente du gradient. Nous pr{\'{e}}sentons aussi les bases th{\'{e}}oriques qui permettent d{\'{e}}tudier la g{\'{e}}n{\'{e}}ralisation {\`{a}} de nouvelles t{\^{a}}ches dune r{\`{e}}gle dapprentissage dont les param{\`{e}}tres ont {\'{e}}t{\'{e}} estim{\'{e}}s {\`{a}} partir dun certain ensemble de t{\^{a}}ches. Enfin, nous exposons bri{\`{e}}vement les r{\'{e}}sultats dune exp{\'{e}}rience consistant {\`{a}} trouver, par descente du gradient, une r{\`{e}}gle dapprentissage pouvant r{\'{e}}soudre plusieurs t{\^{a}}ches bool{\'{e}}ennes lin{\'{e}}airement et non lin{\'{e}}airement s{\'{e}}parables.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	901 addressfr={N{\^i}es, France},topics={BioRules},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	902 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	903
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	904 @INPROCEEDINGS{bengio:1992:oban,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	905 author = {Bengio, Samy and Bengio, Yoshua and Cloutier, Jocelyn and Gecsei, Jan},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	906 title = {On the Optimization of a Synaptic Learning rule},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	907 booktitle = {Conference on Optimality in Biological and Artificial Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	908 year = {1992},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	909 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/bengio_1992_oban.ps},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	910 abstract = {This paper presents a new approach to neural modeling based on the idea of using an automated method to optimize the parameters of a synaptic learning rule. The synaptic modification rule is considered as a parametric function. This function has local inputs and is the same in many neurons. We can use standard optimization methods to select appropriate parameters for a given type of task. We also present a theoretical analysis permitting to study the generalization property of such parametric learning rules. By generalization, we mean the possibility for the learning rule to learn to solve new tasks. Experiments were performed on three types of problems: a biologically inspired circuit (for conditioning in Aplysia). Boolean functions (linearly separable as well as non linearly separable) and classification tasks. The neural network architecture as well as the form and initial parameter values of the synaptic learning function can be designed using a priori knowledge.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	911 addressfr={Dallas, USA},topics={BioRules},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	912 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	913
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	914 @INPROCEEDINGS{bengio:1992:snowbird,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	915 author = {Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	916 title = {Representations Based on Articulatory Dynamics for Speech Recognition},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	917 booktitle = {1992 Neural Networks for Computing Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	918 year = {1992},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	919 topics={PriorKnowledge,Speech},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	920 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	921
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	922 @INPROCEEDINGS{bengio:1993:icann,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	923 author = {Bengio, Samy and Bengio, Yoshua and Cloutier, Jocelyn and Gecsei, Jan},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	924 editor = {Gielen, S. and Kappen, B.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	925 title = {Generalization of a Parametric Learning Rule},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	926 booktitle = {{ICANN} '93: Proceedings of the International Conference on Artificial Neural Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	927 year = {1993},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	928 pages = {502},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	929 publisher = {Springer-Verlag},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	930 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/bengio_1993_icann.ps},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	931 abstract = {In previous work ([4,2,1]) we discussed the subject of parametric learning rules for neural networks. In this article, we present a theoretical basis permitting to study the generalization property of a learning rule whose parameters are estimated from a set of learning tasks. By generalization, we mean the possibility of using the learning rule to learn solve new tasks. Finally, we describe simple experiments on two-dimensional categorization tasks and show how they corroborate the theoretical results.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	932 addressfr={Amsterdam, Pays-Bas},topics={BioRules},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	933 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	934
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	935 @INPROCEEDINGS{bengio:1993:snowbird,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	936 author = {Bengio, Yoshua and Simard, Patrice and Frasconi, Paolo},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	937 title = {The Problem of Learning Long-Term Dependencies in Recurrent Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	938 booktitle = {1993 Neural Networks for Computing Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	939 year = {1993},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	940 topics={LongTerm},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	941 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	942
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	943 @TECHREPORT{bengio:1994,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	944 author = {Bengio, Yoshua and Frasconi, Paolo},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	945 title = {An {EM} Approach to Learning Sequential Behavior},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	946 number = {DSI 11-94},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	947 year = {1994},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	948 institution = {Universita di Firenze, Dipartimento di Sistemi e Informatica},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	949 topics={LongTerm},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	950 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	951
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	952 @INPROCEEDINGS{bengio:1994:acfas,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	953 author = {Bengio, Samy and Bengio, Yoshua and Cloutier, Jocelyn and Gecsei, Jan},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	954 title = {Optimisation d'une r{\`{e}}gle d'apprentissage pour r{\'{e}}seaux de neurones artificiels},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	955 booktitle = {Actes du soixante-deuxi{\`{e}}me congr{\`{e}}s de l'Association Canadienne Fran{\c c}aise pour l'Avancement des Sciences, colloque sur l'apprentissage et les r{\'{e}}seaux de neurones artificiels},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	956 year = {1994},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	957 topics={BioRules},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	958 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	959
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	960 @INPROCEEDINGS{bengio:1994:snowbird,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	961 author = {Bengio, Yoshua and Frasconi, Paolo},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	962 title = {An {EM} Algorithm for Target Propagation},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	963 booktitle = {1994 Neural Networks for Computing Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	964 year = {1994},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	965 topics={LongTerm},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	966 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	967
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	968 @INPROCEEDINGS{bengio:1994:wcci,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	969 author = {Bengio, Samy and Bengio, Yoshua and Cloutier, Jocelyn},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	970 title = {Use of Genetic Programming for the Search of a New Learning Rule for Neural Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	971 booktitle = {Proceedings of the First Conference on Evolutionary Computation, {IEEE} World Congress on Computational Intelligence},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	972 year = {1994},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	973 pages = {324--327},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	974 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/bengio_1994_wcci.ps},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	975 abstract = {In previous work ([1,2,3]), we explained how to use standard optimization methods such as simulated annealing, gradient descent and genetic algorithms to optimize a parametric function which could be used as a learning rule for neural networks. To use these methods, we had to choose a fixed number of parameters and a rigid form for the learning rule. In this article, we propose to use genetic programming to find not only the values of rule parameters but also the optimal number of parameters and the form of the rule. Experiments on classification tasks suggest genetic programming finds better learning rules than other optimization methods. Furthermore, the best rule found with genetic programming outperformed the well-known backpropagation algorithm for a given set of tasks.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	976 topics={BioRules},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	977 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	978
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	979 @INPROCEEDINGS{bengio:1994b:acfas,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	980 author = {Bengio, Yoshua and Frasconi, Paolo},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	981 title = {R{\'{e}}seaux de neurones {M}arkoviens pour l'inf{\'{e}}rence grammaticale},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	982 booktitle = {Actes du soixante-deuxi{\`{e}}me congr{\`{e}}s de l'Association Canadienne Fran{\c c}aise pour l'Avancement des Sciences, colloque sur l'apprentissage et les r{\'{e}}seaux de neurones artificiels},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	983 year = {1994},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	984 topics={Markov,Language},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	985 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	986
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	987 @ARTICLE{bengio:1995:npl,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	988 author = {Bengio, Samy and Bengio, Yoshua and Cloutier, Jocelyn},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	989 title = {On the Search for New Learning Rules for {ANN}s},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	990 journal = {Neural Processing Letters},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	991 volume = {2},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	992 number = {4},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	993 year = {1995},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	994 pages = {26--30},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	995 abstract = {In this paper, we present a framework where a learning rule can be optimized within a parametric learning rule space. We define what we call parametric learning rules and present a theoretical study of their generalization properties when estimated from a set of learning tasks and tested over another set of tasks. We corroborate the results of this study with practical experiments.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	996 topics={BioRules},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	997 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	998
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	999 @INCOLLECTION{bengio:1995:oban,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1000 author = {Bengio, Samy and Bengio, Yoshua and Cloutier, Jocelyn and Gecsei, Jan},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1001 editor = {Levine, D. S. and Elsberry, W. R.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1002 title = {{O}n the Optimization of a Synaptic Learning Rule},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1003 booktitle = {Optimality in Biological and Artificial Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1004 year = {1995},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1005 publisher = {Lawrence Erlbaum},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1006 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/bengio_1995_oban.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1007 abstract = {This paper presents a new approach to neural modeling based on the idea of using an automated method to optimize the parameters of a synaptic learning rule. The synaptic modification rule is considered as a parametric function. This function has local inputs and is the same in many neurons. We can use standard optimization methods to select appropriate parameters for a given type of task. We also present a theoretical analysis permitting to study the generalization property of such parametric learning rules. By generalization, we mean the possibility for the learning rule to learn to solve new tasks. Experiments were performed on three types of problems: a biologically inspired circuit (for conditioning in Aplysia), Boolean functions (linearly separable as well as non linearly separable) and classification tasks. The neural network architecture as well as the form and initial parameter values of the synaptic learning function can be designed using a priori knowledge.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1008 topics={BioRules},cat={B},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1009 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1010
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1011 @TECHREPORT{bengio:1996:udem,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1012 author = {Bengio, Yoshua and Bengio, Samy},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1013 title = {Training Asynchronous Input/Output Hidden {M}arkov Models},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1014 number = {1013},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1015 year = {1996},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1016 institution = {D{\'{e}}partement d'Informatique et de Recherche Op{\'{e}}rationnelle, Universit{\'{e}}de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1017 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/bengio_1996_udem.ps},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1018 topics={Markov},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1019 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1020
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1021 @INPROCEEDINGS{bengio:1997:snowbird,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1022 author = {Bengio, Yoshua and Bengio, Samy and Singer, Yoram and Isabelle, Jean-Fran{\c c}ois},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1023 title = {On the Clusterization of Probabilistic Transducers},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1024 booktitle = {1997 Neural Networks for Computing Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1025 year = {1997},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1026 topics={HighDimensional},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1027 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1028
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1029 @INPROCEEDINGS{bengio:1998:snowbird,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1030 author = {Bengio, Samy and Bengio, Yoshua and Robert, Jacques and B{\'{e}}langer, Gilles},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1031 title = {Stochastic Learning of Strategic Equilibria for Auctions},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1032 booktitle = {Learning Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1033 year = {1998},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1034 topics={Auction},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1035 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1036
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1037 @TECHREPORT{bengio:1998:udem,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1038 author = {Bengio, Samy and Bengio, Yoshua and Robert, Jacques and B{\'{e}}langer, Gilles},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1039 title = {Stochastic Learning of Strategic Equilibria for Auctions},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1040 number = {1119},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1041 year = {1998},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1042 institution = {D{\'{e}}partement d'Informatique et de Recherche Op{\'{e}}rationnelle, Universit{\'{e}}de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1043 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/bengio_1998_udem.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1044 abstract = {This paper presents a new application of stochastic adaptive learning algorithms to the computation of strategic equilibria in auctions. The proposed approach addresses the problems of tracking a moving target and balancing exploration (of action space) versus exploitation (of better modeled regions of action space). Neural networks are used to represent a stochastic decision model for each bidder. Experiments confirm the correctness and usefulness of the approach.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1045 topics={Auction},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1046 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1047
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1048 @INPROCEEDINGS{bengio:1999:snowbird,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1049 author = {Bengio, Yoshua and Latendresse, Simon and Dugas, Charles},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1050 title = {Gradient-Based Learning of Hyper-Parameters},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1051 booktitle = {Learning Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1052 year = {1999},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1053 topics={ModelSelection},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1054 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1055
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1056 @INPROCEEDINGS{bengio:1999:titration,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1057 author = {Bengio, Yoshua and Brault, J-J. and Major, Fran{\c c}ois and Neal, R. and Pigeon, Steven},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1058 title = {Learning Algorithms for Sorting Compounds from Titration Curves},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1059 booktitle = {Symposium on New Perspectives for Computer-Aided Drug Design},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1060 year = {1999},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1061 topics={Speech},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1062 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1063
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1064 @ARTICLE{bengio:2000:ieeetrnn,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1065 author = {Bengio, Samy and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1066 title = {Taking on the Curse of Dimensionality in Joint Distributions Using Neural Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1067 journal = {IEEE Transaction on Neural Networks special issue on data mining and knowledge discovery},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1068 volume = {11},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1069 number = {3},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1070 year = {2000},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1071 pages = {550--557},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1072 abstract = {The curse of dimensionality is severe when modeling high-dimensional discrete data: the number of possible combinations of the variables explodes exponentially. In this paper we propose a new architecture for modeling high-dimensional data that requires resources (parameters and computations) that grow at most as the square of the number of variables, using a multi_layer neural network to represent the joint distribution of the variables as the product of conditional distributions. The neural network can be interpreted as a graphical model without hidden random variables, but in which the conditional distributions are tied through the hidden units. The connectivity of the neural network can be pruned by using dependency tests between the variables (thus reducing significantly the number of parameters). Experiments on modeling the distribution of several discrete data sets show statistically significant improvements over other methods such as naive Bayes and comparable Bayesian networks, and show that significant improvements can be obtained by pruning the network.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1073 topics={HighDimensional,Unsupervised,Mining},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1074 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1075
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1076 @INPROCEEDINGS{bengio:2000:nips,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1077 author = {Bengio, Yoshua and Bengio, Samy},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1078 title = {Modeling High-Dimensional Discrete Data with Multi-Layer Neural Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1079 year = {2000},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1080 pages = {400--406},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1081 crossref = {NIPS12-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1082 abstract = {The curse of dimensionality is severe when modeling high-dimensional discrete data: the number of possible combinations of the variables explodes exponentially. In this paper we propose a new architecture for modeling high-dimensional data that requires resources (parameters and computations) that grow only at most as the square of the number of variables, using a multi-layer neural network to represent the joint distribution of the variables as the product of conditional distributions. The neural network can be interpreted as a graphical model without hidden random variables, but in which the conditional distributions are tied through the hidden units. The connectivity of the neural network can be pruned by using dependency tests between the variables. Experiments on modeling the distribution of several discrete data sets show statistically significant improvements over other methods such as naive Bayes and comparable Bayesian networks, and show that significant improvements can be obtained by pruning the network.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1083 topics={HighDimensional,Unsupervised},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1084 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1085
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1086 @ARTICLE{bengio:2003,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1087 author = {Bengio, Yoshua and Ducharme, R{\'{e}}jean and Vincent, Pascal and Jauvin, Christian},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1088 title = {A Neural Probabilistic Language Model},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1089 volume = {3},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1090 year = {2003},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1091 pages = {1137--1155},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1092 journal = {Journal of Machine Learning Research},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1093 abstract = {A goal of statistical language modeling is to learn the joint probability function of sequences of words in a language. This is intrinsically difficult because of the curse of dimensionality: a word sequence on which the model will be tested is likely to be different from all the word sequences seen during training. Traditional but very successful approaches based on n-grams obtain generalization by concatenating very short overlapping sequences seen in the training set. We propose to fight the curse of dimensionality by learning a distributed representation for words which allows each training sentence to inform the model about an exponential number of semantically neighboring sentences. The model learns simultaneously (1) a distributed representation for each word along with (2) the probability function for word sequences, expressed in terms of these representations. Generalization is obtained because a sequence of words that has never been seen before gets high probability if it is made of words that are similar (in the sense of having a nearby representation) to words forming an already seen sentence. Training such large models (with millions of parameters) within a reasonable time is itself a significant challenge. We report on experiments using neural networks for the probability function, showing on two text corpora that the proposed approach significantly improves on state-of-the-art n-gram models, and that the proposed approach allows to take advantage of longer contexts.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1094 topics={Markov,Unsupervised,Language},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1095 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1096
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1097 @TECHREPORT{bengio:socs-1990,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1098 author = {Bengio, Yoshua and De Mori, Renato and Flammia, Giovanni and Kompe, Ralf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1099 title = {Global Optimization of a Neural Network - Hidden {M}arkov Model Hybrid},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1100 number = {TR-SOCS-90.22},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1101 year = {1990},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1102 institution = {School of Computer Science, McGill University},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1103 topics={Markov},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1104 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1105
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1106 @INPROCEEDINGS{bengioc:1994:acfas,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1107 author = {Bengio, Yoshua and {LeCun}, Yann},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1108 title = {Reconnaissance de mots manuscrits avec r{\'{e}}seaux de neurones et mod{\`{e}}les de {M}arkov},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1109 booktitle = {Actes du soixante-deuxi{\`{e}}me congr{\`{e}}s de l'Association Canadienne Fran{\c c}aise pour l'Avancement des Sciences, colloque sur l'apprentissage et les r{\'{e}}seaux de neurones artificiels},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1110 year = {1994},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1111 topics={Markov,Speech},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1112 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1113
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1114 @TECHREPORT{Bengio_Bottou92,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1115 author = {Bengio, Yoshua and Bottou, {L{\'{e}}on}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1116 title = {A New Approach to Estimating Probability Density Functions with Artificial Neural Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1117 number = {TR-92.02},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1118 year = {1992},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1119 institution = {Massachusetts Institute of Technology, Dept. Brain and Cognitive Sciences},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1120 topics={HighDimensional},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1121 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1122
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1123 @INCOLLECTION{bengio_extension_nips_2003,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1124 author = {Bengio, Yoshua and Paiement, Jean-Fran{\c c}ois and Vincent, Pascal and Delalleau, Olivier and Le Roux, Nicolas and Ouimet, Marie},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1125 keywords = {dimensionality reduction, eigenfunctions learning, Isomap, kernel {PCA}, locally linear embedding, Nystrom formula, spectral methods},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1126 title = {Out-of-Sample Extensions for {LLE}, Isomap, {MDS}, Eigenmaps, and Spectral Clustering},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1127 year = {2004},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1128 crossref = {NIPS16-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1129 abstract = {Several unsupervised learning algorithms based on an eigendecomposition provide either an embedding or a clustering only for given training points, with no straightforward extension for out-of-sample examples short of recomputing eigenvectors. This paper provides a unified framework for extending Local Linear Embedding ({LLE}), Isomap, Laplacian Eigenmaps, Multi-Dimensional Scaling (for dimensionality reduction) as well as for Spectral Clustering. This framework is based on seeing these algorithms as learning eigenfunctions of a data-dependent kernel. Numerical experiments show that the generalizations performed have a level of error comparable to the variability of the embedding algorithms due to the choice of training data.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1130 topics={HighDimensional,Kernel,Unsupervised},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1131 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1132
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1133 @ARTICLE{Bengio_Gingras98a,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1134 author = {Bengio, Yoshua and Gingras, Fran{\c c}ois and Goulard, Bernard and Lina, Jean-Marc},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1135 title = {Gaussian Mixture Densities for Classification of Nuclear Power Plant Data},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1136 journal = {Computers and Artificial Intelligence},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1137 volume = {17},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1138 number = {2-3},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1139 year = {1998},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1140 pages = {189--209},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1141 abstract = {In this paper we are concerned with the application of learning algorithms to the classification of reactor states in nuclear plants. Two aspects must be considered, (1) some types of events (e.g., abnormal or rare) will not appear in the data set, but the system should be able to detect them, (2) not only classification of signals but also their interpretation are important for nuclear plant monitoring. We address both issues with a mixture of mixtures of Gaussians in which some parameters are shared to reflect the similar signals observed in different states of the reactor. An {EM} algorithm for these shared Gaussian mixtures is presented. Experimental results on nuclear plant data demonstrate the advantages of the proposed approach with respect to the above two points.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1142 topics={Mining},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1143 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1144
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1145 @ARTICLE{Bengio_Gingras98b,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1146 author = {Gingras, Fran{\c c}ois and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1147 title = {Handling Asynchronous or Missing Financial Data with Recurrent Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1148 journal = {International Journal of Computational Intelligence and Organizations},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1149 volume = {1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1150 number = {3},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1151 year = {1998},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1152 pages = {154--163},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1153 abstract = {An important issue with many sequential data analysis problems, such as those encountered in financial data sets, is that different variables are known at different frequencies, at different times (asynchronicity), or are sometimes missing. To address this issue we propose to use recurrent networks with feedback into the input units, based on two fundamental ideas. The first motivation is that the filled-in value of the missing variable may not only depend in complicated ways on the value of this variable in the past of the sequence but also on the current and past values of other variables. The second motivation is that, for the purpose of making predictions or taking decisions, it is not always necessary to fill in the best possible value of the missing variables. In fact, it is sufficient to fill in a value which helps the system make better predictions or decisions. The advantages of this approach are demonstrated through experiments on several tasks.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1154 topics={Finance,Missing},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1155 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1156
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1157 @INPROCEEDINGS{Bengio_icassp90,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1158 author = {Bengio, Yoshua and Cardin, Regis and De Mori, Renato and Normandin, Yves},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1159 title = {A Hybrid Coder for Hidden {M}arkov Models Using a Recurrent Neural Network},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1160 booktitle = {International Conference on Acoustics, Speech and Signal Processing},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1161 year = {1990},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1162 pages = {537--540},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1163 topics={Markov,Speech},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1164 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1165
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1166 @INPROCEEDINGS{Bengio_LeCun94,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1167 author = {Bengio, Yoshua and {LeCun}, Yann and Henderson, Donnie},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1168 title = {Globally Trained Handwritten Word Recognizer using Spatial Representation, Space Displacement Neural Networks and Hidden {M}arkov Models},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1169 year = {1994},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1170 pages = {937--944},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1171 crossref = {NIPS6-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1172 abstract = {We introduce a new approach for on-line recognition of handwritten words written in unconstrained mixed style. The preprocessor performs a word-level normalization by fitting a model of the word structure using the {EM} algorithm. Words are then coded into low resolution annotated images where each pixel contains information about trajectory direction and curvature. The recognizer is a convolution network which can be spatially replicated. From the network output, a hidden {Markov} model produces word scores. The entire system is globally trained to minimize word-level errors.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1173 topics={Speech},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1174 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1175
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1176 @ARTICLE{Bengio_LeCun95,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1177 author = {Bengio, Yoshua and {LeCun}, Yann and Nohl, Craig and Burges, Chris},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1178 title = {LeRec: A {NN}/{HMM} Hybrid for On-Line Handwriting Recognition},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1179 journal = {Neural Computation},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1180 volume = {7},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1181 number = {6},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1182 year = {1995},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1183 pages = {1289--1303},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1184 abstract = {We introduce a new approach for on-line recognition of handwritten words written in unconstrained mixed style. The preprocessor performs a word-level normalization by fitting a model of the word structure using the {EM} algorithm. Words are then coded into low resolution annotated images where each pixel contains information about trajectory direction and curvature. The recognizer is a convolution network which can be spatially replicated. From the network output, a hidden {Markov} model produces word scores. The entire system is globally trained to minimize word-level errors.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1185 topics={PriorKnowledge,Speech},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1186 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1187
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1188 @ARTICLE{Bengio_prel92,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1189 author = {Bengio, Yoshua and Gori, Marco and De Mori, Renato},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1190 title = {Learning the Dynamic Nature of Speech with Back-propagation for Sequences},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1191 journal = {Pattern Recognition Letters},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1192 volume = {13},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1193 number = {5},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1194 year = {1992},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1195 pages = {375--385},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1196 note = {(Special issue on Artificial Neural Networks)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1197 topics={Speech},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1198 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1199
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1200 @ARTICLE{Bengio_trnn92,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1201 author = {Bengio, Yoshua and De Mori, Renato and Flammia, Giovanni and Kompe, Ralf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1202 title = {Global Optimization of a Neural Network-Hidden {M}arkov Model Hybrid},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1203 journal = {IEEE Transactions on Neural Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1204 volume = {3},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1205 number = {2},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1206 year = {1992},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1207 pages = {252--259},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1208 topics={Markov},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1209 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1210
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1211 @TECHREPORT{Bergstra+2009,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1212 author = {Bergstra, James and Desjardins, Guillaume and Lamblin, Pascal and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1213 title = {Quadratic Polynomials Learn Better Image Features},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1214 number = {1337},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1215 year = {2009},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1216 institution = {D{\'{e}}partement d'Informatique et de Recherche Op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1217 abstract = {The affine-sigmoidal hidden unit (of the form $\sigma(ax+b)$)
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1218 is a crude predictor of neuron response in visual area V1.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1219 More descriptive models of V1 have been advanced that are no more computationally expensive,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1220 yet artificial neural network research continues to focus on networks of affine-sigmoidal models.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1221 This paper identifies two qualitative differences between the affine-sigmoidal hidden unit
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1222 and a particular recent model of V1 response:
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1223 a) the presence of a low-rank quadratic term in the argument to $\sigma$,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1224 and b) the use of a gentler non-linearity than the $\tanh$ or logistic sigmoid.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1225 We evaluate these model ingredients by training single-layer
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1226 neural networks to solve three image classification tasks.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1227 We experimented with fully-connected hidden units,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1228 as well as locally-connected units and convolutional units
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1229 that more closely mimic the function and connectivity of the visual system.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1230 On all three tasks, both the quadratic interactions and the gentler non-linearity
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1231 lead to significantly better generalization.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1232 The advantage of quadratic units was strongest in conjunction with sparse and convolutional hidden units.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1233 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1234
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1235 @MISC{bergstra+al:2010-scipy,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1236 author = {Bergstra, James},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1237 title = {Optimized Symbolic Expressions and {GPU} Metaprogramming with Theano},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1238 year = {2010},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1239 howpublished = {{SciPy}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1240 note = {Oral}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1241 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1242
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1243 @MISC{bergstra+al:2010-sharcnet,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1244 author = {Bergstra, James and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1245 title = {{GPU} Programming with Theano},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1246 year = {2010},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1247 howpublished = {{SHARCNET} Research Day},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1248 note = {Oral}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1249 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1250
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1251 @MISC{bergstra+al:2010snowbird,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1252 author = {Bergstra, James and Breuleux, Olivier and Bastien, Fr{\'{e}}d{\'{e}}ric and Lamblin, Pascal and Turian, Joseph and Desjardins, Guillaume and Pascanu, Razvan and Erhan, Dumitru and Delalleau, Olivier and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1253 title = {Deep Learning on {GPU}s with Theano},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1254 booktitle = {The Learning Workshop},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1255 year = {2010},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1256 note = {Oral}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1257 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1258
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1259 @INPROCEEDINGS{Bergstra+Bengio-2009,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1260 author = {Bergstra, James and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1261 title = {Slow, Decorrelated Features for Pretraining Complex Cell-like Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1262 year = {2009},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1263 crossref = {NIPS22}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1264 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1265
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1266 @ARTICLE{bergstra+casagrande+erhan+eck+kegl:2006,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1267 author = {Bergstra, James and Casagrande, Norman and Erhan, Dumitru and Eck, Douglas and K{\'{e}}gl, Bal{\'{a}}zs},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1268 title = {Aggregate Features and AdaBoost for Music Classification},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1269 journal = {Machine Learning},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1270 volume = {65},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1271 year = {2006},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1272 pages = {473--484},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1273 issn = {0885-6125},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1274 abstract = {We present an algorithm that predicts musical genre and artist from an audio waveform. Our method uses the ensemble learner ADABOOST to select from a set of audio features that have been extracted from segmented audio and then aggregated. Our classifier proved to be the most effective method for genre classification at the recent MIREX 2005 international contests in music information extraction, and the second-best method for recognizing artists. This paper describes our method in detail, from feature extraction to song classification, and presents an evaluation of our method on three genre databases and two artist-recognition databases. Furthermore, we present evidence collected from a variety of popular features and classifiers that the technique of classifying features aggregated over segments of audio is better than classifying either entire songs or individual short-timescale features.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1275 PDF = {papers/2006_ml_draft.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1276 SOURCE = {OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1277 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1278
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1279 @INPROCEEDINGS{bergstra+lacoste+eck:2006,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1280 author = {Bergstra, James and Lacoste, Alexandre and Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1281 title = {Predicting Genre Labels for Artists using FreeDB},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1282 booktitle = {Proc. 7th International Conference on Music Information Retrieval (ISMIR)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1283 year = {2006},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1284 SOURCE = {OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1285 PDF = {papers/2006_ismir_freedb.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1286 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1287
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1288 @INPROCEEDINGS{bergstra+mandel+eck:2010,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1289 author = {Bergstra, James and Mandel, Michael and Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1290 title = {Scalable Genre and Tag Prediction with Spectral Covariance},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1291 booktitle = {{ISMIR}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1292 year = {2010},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1293 note = {accepted}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1294 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1295
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1296 @MASTERSTHESIS{Bergstra-Msc-2006,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1297 author = {Bergstra, James},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1298 keywords = {apprentissage statistique, classification de musique par genre, extraction de caract{\'{e}}ristiques sonores, recherche d'information musicale},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1299 title = {Algorithms for Classifying Recorded Music by Genre},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1300 year = {2006},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1301 school = {Universit{\'{e}} de Montreal},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1302 abstract = {Ce m{\'{e}}moire traite le probl{\`{e}}me de la classification automatique de signaux musicaux par genre. Dans un premier temps, je pr{\'{e}}sente une technique utilisant l'apprentissage machine pour classifier des statistiques extraites sur des segments du signal sonore. Malgr{\'{e}} le fait que cette technique a d{\'{e}}j{\`{a}} {\'{e}}t{\'{e}} explor{\'{e}}e, mon m{\'{e}}moire est le premier {\`{a}} investiguer l'influence de la longueur et de la quantit{\'{e}} de ces segments sur le taux de classification. J'explore {\'{e}}galement l'importance d'avoir des segments contigus dans le temps. Les segments d'une {\`{a}} trois secondes apportent une meilleure performance, mais pour ce faire, ils doivent {\^{e}}tre suffisamment nombreux. Il peut m{\^{e}}me {\^{e}}tre utile d'augmenter la quantit{\'{e}} de segments jusqu'{\`{a}} ce qu'ils se chevauchent. Dans les m{\^{e}}mes exp{\'{e}}riences, je pr{\'{e}}sente une formulation alternative des descripteurs d'audio nomm{\'{e}}e Melfrequency Cepstral Coefficient (MFCC) qui am{\`{e}}ne un taux de classification de 81 \% sur un jeux de donn{\'{e}}es pour lequel la meilleure performance publi{\'{e}}e est de 71 \%. Cette m{\'{e}}thode de segmentation des chansons, ainsi que cette formulation alternative, ont pour but d'am{\'{e}}liorer l'algorithme gagnant du concours de classification de genre de MIREX 2005, d{\'{e}}velopp{\'{e}} par Norman Casagrande et moi. Ces exp{\'{e}}riences sont un approfondissement du travail entam{\'{e}} par Bergstra et al. [2006a], qui d{\'{e}}crit l'algorithme gagnant de ce concours.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1303 Dans un deuxi{\`{e}}me temps, je pr{\'{e}}sent une m{\'{e}}thode qui utilise FreeDB, une base de donn{\'{e}}es d'information sur les albums, pour attribuer {\`{a}} un artiste une distribution de probabilit{\'{e}} sur son genre. Avec une petite base de donn{\'{e}}es, faite {\`{a}} la main, je montre qu'il y a une haute corr{\'{e}}lation entre cette distribution et l'{\'{e}}tiquette de genre traditionnel. Bien qu'il reste {\`{a}} d{\'{e}}montrer que cette m{\'{e}}thode est utile pour organiser une collection de musique, ce r{\'{e}}sultat sugg{\`{e}}re qu'on peut maintenant {\'{e}}tiqueter de grandes bases de musique automatiquement {\`{a}} un faible co{\^{u}}t, et par cons{\'{e}}quent de poursuivre plus facilement la recherche en classification {\`{a}} grande {\'{e}}chelle. Ce travail sera publi{\'{e}} comme Bergstra et al. [2006b] {\`{a}} ISMIR 2006.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1304 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1305
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1306 @INPROCEEDINGS{bergstra:2010cosyne,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1307 author = {Bergstra, James and Bengio, Yoshua and Lamblin, Pascal and Desjardins, Guillaume and Louradour, Jerome},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1308 title = {Image classification with complex cell neural networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1309 booktitle = {Computational and systems neuroscience (COSYNE)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1310 year = {2010},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1311 note = {Poster},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1312 url = {http://www.frontiersin.org/conferences/individual_abstract_listing.php?conferid=770&pap=3626&ind_abs=1&pg=335},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1313 doi = {10.3389/conf.fnins.2010.03.00334}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1314 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1315
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1316 @INPROCEEDINGS{biaslearn:2000:ijcnn,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1317 author = {Ghosn, Joumana and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1318 title = {Bias Learning, Knowledge Sharing},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1319 booktitle = {International Joint Conference on Neural Networks 2000},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1320 volume = {I},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1321 year = {2000},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1322 pages = {9--14},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1323 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/ijcnn_manifold.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1324 abstract = {Biasing the hypothesis space of a learner has been shown to improve generalisation performances. Methods for achieving this goal have been proposed, that range from deriving and introducing a bias into a learner to automatically learning the bias. In the latter case, most methods learn the bias by simultaneously training several related tasks derived from the same domain and imposing constraints on their parameters. We extend some of the ideas presented in this field and describe a new model that parameterizes the parameters of each task as a function of an affine manifold defined in parameter space and a point lying on the manifold. An analysis of variance on a class of learning tasks is performed that shows some significantly improved performances when using the model.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1325 topics={MultiTask},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1326 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1327
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1328 @ARTICLE{biaslearn:2003:tnn,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1329 author = {Ghosn, Joumana and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1330 title = {Bias Learning, Knowledge Sharing},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1331 journal = {IEEE Transaction on Neural Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1332 volume = {14},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1333 number = {4},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1334 year = {2003},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1335 pages = {748--765},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1336 abstract = {Biasing properly the hypothesis space of a learner has been shown to improve generalization performance. Methods for achieving this goal have been proposed, that range from designing and introducing a bias into a learner to automatically learning the bias. Multitask learning methods fall into the latter category. When several related tasks derived from the same domain are available, these methods use the domain-related knowledge coded in the training examples of all the tasks as a source of bias. We extend some of the ideas presented in this field and describe a new approach that identifies a family of hypotheses, represented by a manifold in hypothesis space, that embodies domain-related knowledge. This family is learned using training examples sampled from a group of related tasks. Learning models trained on these tasks are only allowed to select hypotheses that belong to the family. We show that the new approach encompasses a large variety of families which can be learned. A statistical analysis on a class of related tasks is performed that shows significantly improved performances when using this approach.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1337 topics={MultiTask},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1338 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1339
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1340 @MASTERSTHESIS{Boisvert-Mcs-2005,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1341 author = {Boisvert, Maryse},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1342 keywords = {Algorithme {EM} , D{\'{e}}composition en valeurs singuli{\`{e}}res , D{\'{e}}sambigu{\"{\i}}sation s{\'{e}}mantique , Mod{\`{e}}les graphiques, WordNet },
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1343 title = {R{\'{e}}duction de dimension pour mod{\`{e}}les graphiques probabilistes appliqu{\'{e}}s {\`{a}} la d{\'{e}}sambiguisation s{\'{e}}mantique},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1344 year = {2005},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1345 school = {Universit{\'{e}} de Montr{\'{e}}al}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1346 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1347
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1348 @INPROCEEDINGS{bonneville98,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1349 author = {Bonneville, Martin and Meunier, Jean and Bengio, Yoshua and Soucy, Jean-Paul},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1350 title = {Support Vector Machines for Improving the classification of Brain Pet Images},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1351 booktitle = {SPIE Medical Imaging},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1352 year = {1998},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1353 topics={Kernel},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1354 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1355
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1356 @INPROCEEDINGS{Bottou+Bengio95,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1357 author = {Bottou, {L{\'{e}}on} and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1358 title = {Convergence Properties of the {K}-Means Algorithm},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1359 year = {1995},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1360 pages = {585--592},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1361 crossref = {NIPS7-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1362 abstract = {This paper studies the convergence properties of the well known K-Means clustering algorithm. The K-Means algorithm can be described either as a gradient descent algorithm or by slightly extending the mathematics of the {EM} algorithm to this hard threshold case. We show that the K-Means algorithm actually minimizes the quantization error using the very fast Newton algorithm.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1363 topics={Unsupervised},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1364 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1365
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1366 @ARTICLE{bottou-98,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1367 author = {Bottou, {L{\'{e}}on} and Haffner, Patrick and G. Howard, Paul and Simard, Patrice and Bengio, Yoshua and {LeCun}, Yann},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1368 title = {High Quality Document Image Compression with {DjVu}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1369 journal = {Journal of Electronic Imaging},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1370 volume = {7},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1371 number = {3},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1372 year = {1998},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1373 pages = {410--425},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1374 topics={Compression},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1375 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1376
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1377 @INPROCEEDINGS{Bottou-dcc98,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1378 author = {Bottou, {L{\'{e}}on} and G. Howard, Paul and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1379 editor = {Society, {IEEE} Computer},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1380 title = {The Z-Coder Adaptive Binary Coder},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1381 booktitle = {Data Compression Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1382 year = {1998},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1383 url = {http://leon.bottou.org/papers/bottou-howard-bengio-98},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1384 topics={Compression},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1385 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1386
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1387 @INPROCEEDINGS{bottou-lecun-bengio-97,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1388 author = {Bottou, {L{\'{e}}on} and {LeCun}, Yann and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1389 title = {Global Training of Document Processing Systems using Graph Transformer Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1390 booktitle = {Proc. of Computer Vision and Pattern Recognition},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1391 year = {1997},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1392 pages = {490--494},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1393 publisher = {IEEE},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1394 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/bottou-lecun-bengio-97.ps.gz},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1395 topics={PriorKnowledge,Speech},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1396 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1397
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1398 @TECHREPORT{bottou96TR,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1399 author = {Bottou, {L{\'{e}}on} and Bengio, Yoshua and {LeCun}, Yann},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1400 title = {Document analysis with transducers},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1401 number = {Technical Memorandum HA615600-960701-01TM},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1402 year = {1996},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1403 institution = {AT\&T Labs},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1404 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/transducer-tm.ps.gz},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1405 topics={HighDimensional},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1406 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1407
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1408 @TECHREPORT{bottou97TR,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1409 author = {Bottou, {L{\'{e}}on} and Bengio, Yoshua and G. Howard, Paul},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1410 title = {Z-Coder: A Fast Adaptive Binary Arithmetic Coder},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1411 number = {Technical Memorandum HA615600-970721-02TM},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1412 year = {1997},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1413 institution = {AT\&T Labs},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1414 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/zcoder-tm.ps.gz},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1415 topics={Compression},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1416 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1417
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1418 @MASTERSTHESIS{Bouchard-Msc-2007,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1419 author = {Bouchard, Lysiane},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1420 keywords = {auditory cortex, fMRI, linear classiﬁer, logistic regression, na{\"{\i}}ve bayesian gaussian model, neuroimaging, spectro-temporal modulation, support vectors machine},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1421 title = {Analyse par apprentissage automatique des r{\'{e}}ponses fMRI du cortex auditif {\`{a}} des modulations spectro-temporelles.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1422 year = {2009},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1423 school = {Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1424 abstract = {The application of linear machine learning classiﬁers to the analysis of brain imaging data (fMRI) has led to several interesting breakthroughs in recent years. These classiﬁers combine the responses of the voxels to detect and categorize diﬀerent brain states. They allow a more agnostic analysis than conventional fMRI analysis that systematically treats weak and distributed patterns as unwanted noise. In this project, we use such classiﬁers to validate an hypothesis concerning the encoding of sounds in the human brain. More precisely, we attempt to locate neurons tuned to spectral and temporal modulations in sound. We use fMRI recordings of brain responses of subjects listening to 49 diﬀerent spectro-temporal modulations. The analysis of fMRI data through linear classiﬁers is not yet a standard procedure in this ﬁeld. Thus, an important objective of this project, in the long term, is the development of new machine learning algorithms specialized for neuroimaging data. For these reasons, an important part of the experiments is dedicated to studying the behaviour of the classiﬁers. We are mainly interested in 3 standard linear classiﬁers, namely the support vectors machine algorithm (linear), the logistic regression algorithm (regularized) and the na{\"{\i}}ve bayesian gaussian model (shared variances).}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1425 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1426
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1427 @PHDTHESIS{Boufaden-Phd-2005,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1428 author = {Boufaden, Narj{\`{e}}s},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1429 title = {Extraction d’information {\`{a}} partir de transcriptions de conversations t{\'{e}}l{\'{e}}phoniques sp{\'{e}}cialis{\'{e}}es},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1430 year = {2005},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1431 school = {Universit{\'{e}} de Montr{\'{e}}al, D{\'{e}}partement d'Informatique et de Recherche Op{\'{e}}rationnel}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1432 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1433
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1434 @INPROCEEDINGS{Carreau+Bengio-2007,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1435 author = {Carreau, Julie and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1436 title = {A Hybrid {Pareto} Model for Conditional Density Estimation of Asymmetric Fat-Tail Data},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1437 booktitle = {Proceedings of the Eleventh International Conference on Artificial Intelligence and Statistics (AISTATS'07)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1438 year = {2007},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1439 publisher = {Omnipress},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1440 abstract = {We propose an estimator for the conditional density p(Y\|X) that can adapt for asymmetric heavy tails which might depend on X. Such estimators have important applications in finance and insurance. We draw from Extreme Value Theory the tools to build a hybrid unimodal density having a parameter controlling the heaviness of the upper tail. This hybrid is a Gaussian whose upper tail has been replaced by a generalized {Pareto} tail. We use this hybrid in a multi-modal mixture in order to obtain a nonparametric density estimator that can easily adapt for heavy tailed data. To obtain a conditional density estimator, the parameters of the mixture estimator can be seen as
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1441 functions of X and these functions learned. We show experimentally that this approach better models the conditional density in terms of likelihood than compared competing algorithms : conditional mixture models with other types of components and multivariate nonparametric models.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1442 date={21-24}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1443 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1444
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1445 @ARTICLE{Carreau+Bengio-2009,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1446 author = {Carreau, Julie and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1447 title = {A Hybrid {Pareto} Mixture for Conditional Asymmetric Fat-Tailed Distributio\ n},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1448 journal = {IEEE Transactions on Neural Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1449 volume = {20},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1450 number = {7},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1451 year = {2009},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1452 pages = {1087--1101},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1453 issn = {1045-9227},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1454 abstract = {In many cases, we observe some variables X that contain predictive information over a scalar variable of interest Y, with (X,Y) pairs observed in a training set. We can take advantage of this information to estimate the conditional density P(Y\X = x). In this paper, we propose a conditional mixture model with hybrid {Pareto} components to estimate P(Y\X = x).The hybrid {Pareto} is a Gaussian whose upper tail has been replaced by a generalized {Pareto} tail. A third parameter, in addition to the location and spread parameters of the Gaussian, controls the heaviness of the upper tail. Using the hybrid {Pareto} in a mixture model results in a nonparametric estimator that can adapt to multimodality, asymmetry, and heavy tails. A conditional density estimator is built by modeling the parameters of the mixture estimator as functions of X. We use a neural network to implement these functions. Such conditional density estimators have important applications in many domains such as finance and insurance. We show experimentally that this novel approach better models the conditional density in terms of likelihood, compared to competing algorithms: conditional mixture models with other types of components and a classical kernel-based nonparametric model.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1455 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1456
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1457 @ARTICLE{Carreau+Bengio-extreme-2009,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1458 author = {Carreau, Julie and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1459 title = {A Hybrid {Pareto} Model for Asymmetric Fat-Tailed Data: the univariate case},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1460 journal = {Extremes},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1461 volume = {12},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1462 number = {1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1463 year = {2009},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1464 pages = {53--76},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1465 abstract = {Density estimators that can adapt to asymmetric heavy tails are required in many applications such as finance and insurance. Extreme Value Theory (EVT) has developped principled methods based on asymptotic results to estimate the tails of most distributions. However, the finite sample approximation might introduce a severe bias in many cases. Moreover, the full range of the distribution is often needed, not only the tail area. On the other hand, non-parametric methods, while being powerful where data are abundant, fail to extrapolate properly in the tail area. We put forward a non-parametric density estimator that brings together the strengths of non-parametric density estimation and of EVT. A hybrid {Pareto} distribution that can be used in a mixture model is proposed to extend the generalized {Pareto} (GP) to the whole real axis. Experiments on simulated data show the following. On one hand, the mixture of hybrid {Pareto}s converges faster in terms of log-likelihood and provides good estimates of the tail of the distributions when compared with other density estimators including the GP distribution. On the other hand, the mixture of hybrid {Pareto}s offers an alternate way to estimate the tail index which is comparable to the one estimated with the standard GP methodology. The mixture of hybrids is also evaluated on the Danish fire insurance data set.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1466 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1467
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1468 @PHDTHESIS{Carreau-PhD-2007,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1469 author = {Carreau, Julie},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1470 keywords = {density estimation, extreme values, generalized {Pareto} distribution, heavy-tailed distribution, mixture of distributions, neural networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1471 title = {Mod{\`{e}}les {Pareto} hybrides pour distributions asym{\'{e}}triques et {\`{a}} queues lourdes},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1472 year = {2007},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1473 school = {UdeM},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1474 abstract = {We put forward a class of density estimators that can adapt to asymmetric, multi-modal and heavy-tailed distributions. Such distributions occur in many application domains such as finance and insurance. Mixture of gaussians are flexible non-parametric density estimators that have good approximation properties when the number of components is well chosen with respect to the training set size. However, those models are performing poorly on heavy-tailed data because few observations occur in the tail area. To solve this problem, we resort to extreme value theory where methods based on sound parametric assumptions have been developped to enable extrapolation beyond the range of the observations. More precisely, we build on the PoT method that was developped in hydrology where PoT stands for "Peaks-over-Threshold". The observations exceeding a given threshold are modeled by the generalized {Pareto} distribution. This distribution can approximate arbitrarily well the tail of most distributions. We build a new distribution, the hybrid {Pareto}, by stitching together a truncated Normal and a generalized {Pareto} distribution. We impose continuity constraints at the junction point. The hybrid {Pareto} is thus a smooth distribution that can be used in a mixture model. The behavior of the upper tail of the hybrid is similar to the behavior of the generalized {Pareto} tail. Moreover, the threshold inherent in the the PoT methodology can now be defined implicitly as the junction point of the component with the heaviest tail. This component also determines the tail index of the mixture. Hence, the hybrid {Pareto} mixture offers an alternate way to estimate the tail index associated with heavy-tailed data. In several applications, information that has predictive power on the variable of interest is available. In that case, we want to model the conditional density of Y given X, the vector containing predictive information. When the distribution of Y given X is asymmetric, multi-modal and heavy-tailed, we propose to use a mixure of hybrid {Pareto}s whose parameters are functions of X. Those functions are implemented by means of a neural network with one hidden layer. Neural neworks are non-parametric models that can, in principle, approximate any continuous function. Experiments on artificial and real data sets show that the hybrid {Pareto} mixture, unconditional and conditional, outperforms other density estimators in terms of log-likelihood.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1475 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1476
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1477 @INPROCEEDINGS{casagrande+eck+kegl:icmc2005,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1478 author = {Casagrande, Norman and Eck, Douglas and K{\'{e}}gl, Bal{\'{a}}zs},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1479 title = {Geometry in Sound: A Speech/Music Audio Classifier Inspired by an Image Classifier},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1480 booktitle = {{Proceedings of the International Computer Music Conference (ICMC)}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1481 year = {2005},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1482 pages = {207--210},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1483 url = {http://www.iro.umontreal.ca/~eckdoug/papers/2005_icmc_casagrande.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1484 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1485 sourcetype={Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1486 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1487
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1488 @INPROCEEDINGS{casagrande+eck+kegl:ismir2005,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1489 author = {Casagrande, Norman and Eck, Douglas and K{\'{e}}gl, Bal{\'{a}}zs},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1490 title = {Frame-Level Audio Feature Extraction using {A}da{B}oost},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1491 booktitle = {{Proceedings of the 6th International Conference on Music Information Retrieval ({ISMIR} 2005)}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1492 year = {2005},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1493 pages = {345--350},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1494 url = {http://www.iro.umontreal.ca/~eckdoug/papers/2005_ismir_casagrande.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1495 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1496 sourcetype={Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1497 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1498
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1499 @PROCEEDINGS{ccai2006,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1500 editor = {Lamontagne, Luc and Marchand, Mario},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1501 title = {Advances in Artificial Intelligence, 19th Conference of the Canadian Society for Computational Studies of Intelligence, Canadian AI 2006, Qu{\'{e}}bec City, Qu{\'{e}}bec, Canada, June 7-9, 2006, Proceedings},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1502 booktitle = {Canadian Conference on AI},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1503 series = {Lecture Notes in Computer Science},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1504 volume = {4013},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1505 year = {2006},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1506 publisher = {Springer}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1507 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1508
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1509 @INPROCEEDINGS{Chapados+Bengio-2006,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1510 author = {Chapados, Nicolas and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1511 title = {The K Best-Paths Approach to Approximate Dynamic Programming with Application to Portfolio Optimization},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1512 booktitle = {AI06},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1513 year = {2006},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1514 pages = {491-502}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1515 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1516
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1517 @INPROCEEDINGS{Chapados+Bengio-2007,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1518 author = {Chapados, Nicolas and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1519 title = {Forecasting Commodity Contract Spreads with Gaussian Process},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1520 booktitle = {13th Intarnational Conference on Computing in Economics and Finance},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1521 year = {2007},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1522 abstract = {We introduce a functional representation of time series which allows forecasts to be performed over an unspecified horizon with progressively-revealed information sets. By virtue of using Gaussian processes, a complete covariance matrix between forecasts at several time-steps is available. This information is put to use in an application to actively trade price spreads between commodity futures contracts. The approach delivers impressive out-of-sample risk-adjusted returns after transaction costs on a portfolio of 30 spreads.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1523 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1524
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1525 @ARTICLE{Chapados+Bengio-2008-JOC,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1526 author = {Chapados, Nicolas and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1527 title = {Noisy K Best-Paths for Approximate Dynamic Programming with Application to Portfolio Optimization},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1528 journal = {Journal of Computers},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1529 volume = {2},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1530 number = {1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1531 year = {2007},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1532 pages = {12--19},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1533 abstract = {We describe a general method to transform a non-Markovian sequential decision problem into a supervised learning problem using a K-bestpaths algorithm. We consider an application in financial portfolio management where we can train a controller to directly optimize a Sharpe Ratio (or other risk-averse non-additive) utility function. We illustrate the approach by demonstrating experimental results using a kernel-based controller architecture that would not normally be considered in traditional
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1534 reinforcement learning or approximate dynamic programming.We further show that using a non-additive criterion (incremental Sharpe Ratio) yields a noisy K-best-paths extraction problem, that can give substantially improved performance.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1535 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1536
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1537 @MASTERSTHESIS{Chapados-Msc-2000,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1538 author = {Chapados, Nicolas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1539 title = {Crit{\`{e}}res d'optimisation d'algorithmes d'apprentissage en gestion de portefeuille},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1540 year = {2000},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1541 school = {Universit{\'{e}} de Montr{\'{e}}al}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1542 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1543
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1544 @INPROCEEDINGS{chapados2000,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1545 author = {Chapados, Nicolas and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1546 title = {Cost Functions and Model Combination for {VaR}-Based Asset Allocation Using Neural Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1547 booktitle = {Computational Finance 2000},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1548 year = {2000},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1549 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/compfin2000_final.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1550 abstract = {We introduce an asset-allocation framework based on the active control of the value-at-risk of the portfolio. Within this framework, we compare two paradigms for making the allocation using neural networks. The first one uses the network to make a forecast of asset behavior, in conjunction with a traditional mean-variance allocator for constructing the portfolio. The second paradigm uses the network to directly make the portfolio allocation decisions. We consider a method for performing soft input variable selection, and show its considerable utility. We use model combination (committee) methods to systematize the choice of hyperparemeters during training. We show that committees using both paradigms are significantly outperforming the benchmark market performance.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1551 topics={Finance},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1552 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1553
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1554 @ARTICLE{chapados:2001,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1555 author = {Chapados, Nicolas and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1556 title = {Cost Functions and Model Combination for VaR--based Asset Allocation using Neural Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1557 journal = {IEEE Transactions on Neural Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1558 volume = {12},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1559 number = {4},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1560 year = {2001},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1561 pages = {890--906},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1562 abstract = {We introduce an asset-allocation framework based on the active control of the value-at-risk of the portfolio. Within this framework, we
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1563 compare two paradigms for making the allocation using neural networks. The first one uses the network to make a forecast of asset behavior, in conjunction with a traditional mean-variance allocator for constructing the portfolio. The second paradigm uses the network to directly make the portfolio allocation decisions. We consider a method for performing soft input variable selection, and show its considerable utility. We use model combination (committee) methods to systematize the choice of hyperparemeters during training. We show that committees
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1564 using both paradigms are significantly outperforming the benchmark market performance.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1565 topics={Finance},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1566 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1567
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1568 @ARTICLE{chapados:2003,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1569 author = {Bengio, Yoshua and Chapados, Nicolas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1570 title = {Extensions to Metric-Based Model Selection},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1571 year = {2003},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1572 journal = {Journal of Machine Learning Research},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1573 abstract = {Metric-based methods have recently been introduced for model selection and regularization, often yielding very significant improvements over the alternatives tried (including cross-validation). All these methods require unlabeled data over which to compare functions and detect gross differences in behavior away from the training points. We introduce three new extensions of the metric model selection methods and apply them to feature selection. The first extension takes advantage of the particular case of time-series data in which the task involves prediction with a horizon h. The idea is to use at t the h unlabeled examples that precede t for model selection. The second extension takes advantage of the different error distributions of cross-validation and the metric methods: cross-validation tends to have a larger variance and is unbiased. A hybrid combining the two model selection methods is rarely beaten by any of the two methods. The third extension deals with the case when unlabeled data is not available at all, using an estimated input density. Experiments are described to study these extensions in the context of capacity control and feature subset selection.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1574 topics={ModelSelection,Finance},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1575 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1576
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1577 @ARTICLE{chapelle:2001,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1578 author = {Chapelle, Olivier and Vapnik, Vladimir and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1579 title = {Model Selection for Small Sample Regression},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1580 journal = {Machine Learning},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1581 year = {2001},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1582 abstract = {Model selection is an important ingredient of many machine learning algorithms, in particular when the sample size in small, in order to strike the right trade-off between overfitting and underfitting. Previous classical results for linear regression are based on an asymptotic analysis. We present a new penalization method for performing model selection for regression that is appropriate even for small samples. Our penalization is based on an accurate estimator of the ratio of the expected training error and the expected generalization error, in terms of the expected eigenvalues of the input covariance matrix.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1583 topics={ModelSelection},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1584 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1585
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1586 @INCOLLECTION{chapter-eval-longterm-2001,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1587 author = {Schmidhuber, Juergen and Hochreiter, Sepp and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1588 editor = {Kolen, J. and Kremer, S.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1589 title = {Evaluating Benchmark Problems by Random Guessing},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1590 booktitle = {Field Guide to Dynamical Recurrent Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1591 year = {2001},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1592 publisher = {IEEE Press},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1593 topics={LongTerm},cat={B},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1594 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1595
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1596 @INCOLLECTION{chapter-gradient-document-2001,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1597 author = {{LeCun}, Yann and Bottou, {L{\'{e}}on} and Bengio, Yoshua and Haffner, Patrick},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1598 editor = {Haykin, S. and Kosko, B.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1599 title = {Gradient-Based Learning Applied to Document Recognition},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1600 booktitle = {Intelligent Signal Processing},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1601 year = {2001},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1602 pages = {306--351},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1603 publisher = {IEEE Press},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1604 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/lecun-01a.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1605 abstract = {Multilayer Neural Networks trained with a backprppagation algorithm constitute the best example of a successful Gradient-Based Learning technique. Given an appropriate network architecture, Gradient-Based Learning algorithms can be used to synthesize a complex decision surface that can classify high-dimensional patterns such as handwritten characters, with minimal preprocessing. This paper reviews various methods applied to handwritten character recognition and compares them on a standard handwritten digit recognition task. Convolutional Neural Networks, that are specifically designed to deal with the variability of 2D shapes, are shown to outperform all other techniques.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1606 Real-life document recognition systems are composed of multiple modules including field extraction, segmentation, recognition, and language modeling. A new learning paradigm, called Graph Transformer Networks (GTN), allows such multi-module systems to be trained globally using Gradient-Based methods so as to monimize an overall peformance measure.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1607 Two systems for on-line handwriting recognition are described. Experiments demonstrate the advantage of global training, and the flexibility of Graph Transformer Networks.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1608 A Graph Transformer Network for reading bank check is also described. It uses Convolutional Neural Network character recognizers combined with a global training technique to provides record accuracy on business and personal checks. It is deployed commercially and reads several million checks per day.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1609 topics={PriorKnowledge,Speech},cat={B},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1610 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1611
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1612 @INCOLLECTION{chapter-gradient-flow-2001,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1613 author = {Hochreiter, Sepp and Bengio, Yoshua and Frasconi, Paolo},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1614 editor = {Kolen, J. and Kremer, S.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1615 title = {Gradient Flow in Recurrent Nets: the Difficulty of Learning Long-Term Dependencies},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1616 booktitle = {Field Guide to Dynamical Recurrent Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1617 year = {2001},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1618 publisher = {IEEE Press},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1619 topics={LongTerm},cat={B},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1620 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1621
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1622 @INPROCEEDINGS{chemero+eck:1999,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1623 author = {Chemero, T. and Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1624 title = {An Exploration of Representational Complexity via Coupled Oscillators},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1625 booktitle = {{Proceedings of the Tenth Midwest Artificial Intelligence and Cognitive Science Society}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1626 year = {1999},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1627 publisher = {MIT Press},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1628 url = {http://www.iro.umontreal.ca/~eckdoug/papers/1999_chemero.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1629 abstract = {We note some inconsistencies in a view of representation which takes {\it decoupling} to be of key importance. We explore these inconsistencies using examples of representational vehicles taken from coupled oscillator theory and suggest a new way to reconcile {\it coupling} with {\it absence}. Finally, we tie these views to a teleological definition of representation.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1630 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1631 sourcetype={Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1632 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1633
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1634 @ARTICLE{ChemInfModel2006,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1635 author = {Erhan, Dumitru and {L'Heureux}, Pierre-Jean and Yue, Shi Yi and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1636 title = {Collaborative Filtering on a Family of Biological Targets},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1637 journal = {J. Chem. Inf. Model.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1638 volume = {46},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1639 number = {2},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1640 year = {2006},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1641 pages = {626--635},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1642 abstract = {Building a QSAR model of a new biological target for which few screening data are available is a statistical
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1643 challenge. However, the new target may be part of a bigger family, for which we have more screening data.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1644 Collaborative filtering or, more generally, multi-task learning, is a machine learning approach that improves
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1645 the generalization performance of an algorithm by using information from related tasks as an inductive
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1646 bias. We use collaborative filtering techniques for building predictive models that link multiple targets to
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1647 multiple examples. The more commonalities between the targets, the better the multi-target model that can
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1648 be built. We show an example of a multi-target neural network that can use family information to produce
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1649 a predictive model of an undersampled target. We evaluate JRank, a kernel-based method designed for
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1650 collaborative filtering. We show their performance on compound prioritization for an HTS campaign and
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1651 the underlying shared representation between targets. JRank outperformed the neural network both in the
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1652 single- and multi-target models.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1653 topics={Bioinformatic,MultiTask},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1654 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1655
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1656 @TECHREPORT{collobert:2001:rr01-12,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1657 author = {Collobert, Ronan and Bengio, Samy and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1658 title = {A Parallel Mixture of {SVM}s for Very Large Scale Problems},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1659 number = {12},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1660 year = {2001},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1661 institution = {IDIAP},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1662 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/IDIAP-RR-01-12.ps},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1663 abstract = {Support Vector Machines ({SVM}s) are currently the state-of-the-art models for many classification problems but they suffer from the complexity of their training algorithm which is at least quadratic with respect to the number of examples. Hence, it is hopeless to try to solve real-life problems having more than a few hundreds of thousands examples with {SVM}s. The present paper proposes a new mixture of {SVM}s that can be easily implemented in parallel and where each {SVM} is trained on a small subset of the whole dataset. Experiments on a large benchmark dataset (Forest) yielded significant time improvement (time complexity appears empirically to locally grow linearly with the number of examples). In addition, and that is a surprise, a significant improvement in generalization was observed.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1664 topics={Kernel},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1665 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1666
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1667 @ARTICLE{collobert:2002,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1668 author = {Collobert, Ronan and Bengio, Samy and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1669 title = {Parallel Mixture of {SVM}s for Very Large Scale Problem},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1670 journal = {Neural Computation},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1671 year = {2002},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1672 abstract = {Support Vector Machines ({SVM}s) are currently the state-of-the-art models for many classification problems but they suffer from the complexity of their training algorithm which is at least quadratic with respect to the number of examples. Hence, it is hopeless to try to solve real-life problems having more than a few hundreds of thousands examples with {SVM}s. The present paper proposes a new mixture of {SVM}s that can be easily implemented in parallel and where each {SVM} is trained on a small subset of the whole dataset. Experiments on a large benchmark dataset (Forest) yielded significant time improvement (time complexity appears empirically to locally grow linearly with the number of examples). In addition, and that is a surprise, a significant improvement in generalization was observed.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1673 topics={HighDimensional,Kernel},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1674 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1675
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1676 @BOOK{collobert:2002:book,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1677 author = {Collobert, Ronan and Bengio, Yoshua and Bengio, Samy},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1678 editor = {Lee, S. W. and Verri, A.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1679 title = {Scaling Large Learning Problems with Hard Parallel Mixtures},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1680 booktitle = {Pattern Recognition with Support Vector Machines},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1681 series = {Lecture Notes in Computer Science},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1682 volume = {2388},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1683 year = {2002},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1684 publisher = {Springer-Verlag},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1685 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/2002_mixtures_svm.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1686 abstract = {A challenge for statistical learning is to deal with large data sets, e.g. in data mining. Popular learning algorithms such as Support Vector Machines have training time at least quadratic in the number of examples: they are hopeless to solve prolems with a million examples. We propose a "hard parallelizable mixture" methodology which yields significantly reduced training time through modularization and parallelization: the training data is iteratively partitioned by a "gater" model in such a way that it becoms easy to learn an "expert" model separately in each region of the parition. A probabilistic extension and the use of a set of generative models allows representing a gater so that all pieces of the model are locally trained. For {SVM}s, time complexity appears empirically to locally grow linearly with the number of examples, while generalization performance can be enhanced. For the probabilistic version of the algorithm, the iterative algorithm provably goes down in a cost function that is an upper bound on the negative log-likelihood.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1687 topics={Kernel},cat={B},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1688 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1689
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1690 @MISC{copyright-CTAI,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1691 author = {Bengio, Yoshua and Ducharme, R{\'{e}}jean and Dorion, Christian},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1692 title = {Commodity Trading Advisor Index},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1693 year = {2004-2009},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1694 howpublished = {copyright, and commercialized software license.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1695 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1696
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1697 @MISC{copyright-PLearn,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1698 author = {Vincent, Pascal and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1699 title = {{PLearn}, a {C++} Machine Learning Library},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1700 year = {1998-2009},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1701 howpublished = {copyright, public domain license.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1702 url = {www.plearn.org}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1703 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1704
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1705 @ARTICLE{Cosi90,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1706 author = {Cosi, Piero and Bengio, Yoshua and De Mori, Renato},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1707 title = {Phonetically-based multi-layered networks for acoustic property extraction and automatic speech recognition},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1708 journal = {Speech Communication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1709 volume = {9},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1710 number = {1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1711 year = {1990},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1712 pages = {15--30},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1713 topics={PriorKnowledge,Speech},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1714 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1715
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1716 @INCOLLECTION{courville+eck+bengio:nips2009,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1717 author = {Courville, Aaron and Eck, Douglas and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1718 editor = {},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1719 title = {An Infinite Factor Model Hierarchy Via a Noisy-Or Mechanism},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1720 booktitle = {Neural Information Processing Systems Conference (NIPS) 22},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1721 year = {2009},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1722 pages = {405--413},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1723 publisher = {},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1724 url = {http://books.nips.cc/papers/files/nips22/NIPS2009_1100.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1725 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1726 sourcetype={Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1727 pdf={""},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1728 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1729
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1730 @INPROCEEDINGS{davies+plumbley+eck:waspaa2009,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1731 author = {Davies, M. and Plumbley, M. and Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1732 title = {Towards a musical beat emphasis function},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1733 booktitle = {Proceedings of IEEE WASPAA},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1734 year = {2009},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1735 organization = {IEEE Workshop on Applications of Signal Processing to Audio and Acoustics},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1736 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1737 sourcetype={Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1738 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1739
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1740 @INPROCEEDINGS{Delalleau+al-2005,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1741 author = {Delalleau, Olivier and Bengio, Yoshua and Le Roux, Nicolas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1742 editor = {Cowell, Robert G. and Ghahramani, Zoubin},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1743 title = {Efficient Non-Parametric Function Induction in Semi-Supervised Learning},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1744 booktitle = {Proceedings of the Tenth International Workshop on Artificial Intelligence and Statistics (AISTATS'05)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1745 year = {2005},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1746 pages = {96--103},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1747 publisher = {Society for Artificial Intelligence and Statistics},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1748 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/semisup_aistats2005.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1749 abstract = {There has been an increase of interest for semi-supervised learning recently, because of the many datasets with large amounts of unlabeled examples and only a few labeled ones. This paper follows up on proposed nonparametric algorithms which provide an estimated continuous label for the given unlabeled examples. First, it extends them to function induction algorithms that minimize a regularization criterion applied to an out-of-sample example, and happen to have the form of Parzen windows regressors. This allows to predict test labels without solving again a linear system of dimension n (the number of unlabeled and labeled training examples), which can cost O(n^3). Second, this function induction procedure gives rise to an efficient approximation of the training process, reducing the linear system to be solved to m << n unknowns, using only a subset of m examples. An improvement of O(n^2/m^2) in time can thus be obtained. Comparative experiments are presented, showing the good performance of the induction formula and approximation algorithm.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1750 topics={Unsupervised},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1751 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1752
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1753 @INCOLLECTION{Delalleau+al-ssl-2006,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1754 author = {Delalleau, Olivier and Bengio, Yoshua and Le Roux, Nicolas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1755 editor = {Chapelle, Olivier and {Sch{\"{o}}lkopf}, Bernhard and Zien, Alexander},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1756 title = {Large-Scale Algorithms},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1757 booktitle = {Semi-Supervised Learning},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1758 year = {2006},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1759 pages = {333--341},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1760 publisher = {{MIT} Press},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1761 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/delalleau_ssl.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1762 abstract = {In Chapter 11, it is shown how a number of graph-based semi-supervised learning
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1763 algorithms can be seen as the minimization of a specific cost function, leading to a
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1764 linear system with n equations and unknowns (with n the total number of labeled
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1765 and unlabeled examples). Solving such a linear system will in general require on the
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1766 order of O(kn2) time and O(kn) memory (for a sparse graph where each data point
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1767 has k neighbors), which can be prohibitive on large datasets (especially if k = n,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1768 i.e. the graph is dense). We present in this chapter a subset selection method that
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1769 can be used to reduce the original system to one of size m << n. The idea is to solve
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1770 for the labels of a subset S of X of only m points, while still retaining information
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1771 from the rest of the data by approximating their label with a linear combination of
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1772 the labels in S (using the induction formula presented in Chapter 11). This leads
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1773 to an algorithm whose computational requirements scale as O(m2n) and memory
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1774 requirements as O(m2), thus allowing one to take advantage of significantly bigger
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1775 unlabeled datasets than with the original algorithms.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1776 cat={B},topics={Unsupervised},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1777 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1778
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1779 @INCOLLECTION{DeMori90a,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1780 author = {De Mori, Renato and Bengio, Yoshua and Cosi, Piero},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1781 editor = {Mohr, R. and Pavlidis, T. and Sanfelin, A.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1782 title = {On the use of an ear model and multi-layer networks for automatic speech recognition},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1783 booktitle = {Structural Pattern Analysis},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1784 year = {1990},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1785 publisher = {World Scientific},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1786 topics={PriorKnowledge,Speech},cat={B},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1787 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1788
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1789 @INPROCEEDINGS{Desjardins+al-2010,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1790 author = {Desjardins, Guillaume and Courville, Aaron and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1791 title = {Tempered {Markov} Chain Monte Carlo for training of Restricted {Boltzmann} Machine},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1792 booktitle = {Proceedings of AISTATS 2010},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1793 volume = {9},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1794 year = {2010},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1795 pages = {145-152},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1796 abstract = {Alternating Gibbs sampling is the most common scheme used for sampling from Restricted {Boltzmann} Machines (RBM), a crucial component in deep architectures such as Deep Belief Networks. However, we find that it often does a very poor job of rendering the diversity of modes captured by the trained model. We suspect that this hinders the advantage that could in principle be brought by training algorithms relying on Gibbs sampling for uncovering spurious modes, such as the Persistent Contrastive Divergence algorithm. To alleviate this problem, we explore the use of tempered {Markov} Chain Monte-Carlo for sampling in RBMs. We find both through visualization of samples and measures of likelihood on a toy dataset that it helps both sampling and learning.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1797 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1798
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1799 @TECHREPORT{Desjardins-2008,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1800 author = {Desjardins, Guillaume and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1801 keywords = {Convolutional Architectures, Deep Networks, RBM, Vision},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1802 title = {Empirical Evaluation of Convolutional RBMs for Vision},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1803 number = {1327},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1804 year = {2008},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1805 institution = {D{\'{e}}partement d'Informatique et de Recherche Op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1806 abstract = {Convolutional Neural Networks ({CNN}) have had great success in machine learning tasks involving vision and represent one of the early successes of deep networks. Local receptive fields and weight
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1807 sharing make their architecture ideally suited for vision tasks by helping to enforce a prior based on our knowledge of natural images. This same prior could also be applied to recent developments in the field of deep networks, in order to tailor these new architectures for artificial vision. In this context, we show how the Restricted {Boltzmann} Machine (RBM), the building block of Deep Belief Networks (DBN), can be adapted to operate in a convolutional manner. We compare their performance to standard fully-connected RBMs on a simple visual learning task and show that the convolutional RBMs (CRBMs) converge to smaller values of the negative likelihood function. Our experiments also indicate that CRBMs are more efficient than standard RBMs trained on small image patches, with the CRBMs having faster convergence.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1808 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1809
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1810 @TECHREPORT{Desjardins-tech-2009,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1811 author = {Desjardins, Guillaume and Courville, Aaron and Bengio, Yoshua and Vincent, Pascal and Delalleau, Olivier},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1812 keywords = {CD, PCD, RBM, simulated tempering, tempered MCMC, unsupervised learning},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1813 title = {Tempered {Markov} Chain Monte Carlo for training of Restricted {Boltzmann} Machines},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1814 number = {1345},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1815 year = {2009},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1816 institution = {D{\'{e}}partement d'Informatique et de Recherche Op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1817 abstract = {Alternating Gibbs sampling is the most common scheme used for sampling from Restricted {Boltzmann} Machines (RBM), a crucial component in deep architectures such as Deep Belief Networks. However, we find that it often does a very poor job of rendering the diversity of modes captured by the trained model. We suspect that this hinders the advantage that could in principle be brought by training algorithms relying on Gibbs sampling for uncovering spurious modes, such as the Persistent Contrastive Divergence algorithm. To alleviate this problem, we
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1818 explore the use of tempered {Markov} Chain Monte-Carlo for sampling in RBMs. We find both through visualization of samples and measures of likelihood that it helps both sampling and learning.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1819 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1820
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1821 @ARTICLE{Dugas+Bengio-2009,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1822 author = {Dugas, Charles and Bengio, Yoshua and Belisle, Francois and Nadeau, Claude and Garcia, Rene},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1823 title = {Incorporating Functional Knowledge in Neural Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1824 journal = {The Journal of Machine Learning Research},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1825 volume = {10},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1826 year = {2009},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1827 pages = {1239--1262},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1828 abstract = {Incorporating prior knowledge of a particular task into the architecture of a learning algorithm can greatly improve generalization performance. We study here a case where we know that the function to be learned is non-decreasing in its two arguments and convex in one of them. For this purpose we propose a class of functions similar to multi-layer neural networks but (1) that has those properties, (2) is a universal approximator of Lipschitz functions with these and other properties. We apply this new class of functions to the task of modelling the price of call options. Experiments show improvements on regressing the price of call options using the new types of function classes that incorporate the a priori constraints.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1829 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1830
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1831 @PHDTHESIS{Dugas-Phd-2003,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1832 author = {Dugas, Charles},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1833 title = {Les algorithmes d'apprentissage appliqu{\'{e}}s aux risques financiers},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1834 year = {2003},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1835 school = {Universit{\'{e}} de Montr{\'{e}}al}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1836 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1837
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1838 @ARTICLE{dugas:2003,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1839 author = {Dugas, Charles and Bengio, Yoshua and Chapados, Nicolas and Vincent, Pascal and Denoncourt, Germain and Fournier, Christian},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1840 title = {Statistical Learning Algorithms Applied to Automobile Insurance Ratemaking},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1841 journal = {CAS Forum},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1842 volume = {1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1843 number = {1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1844 year = {2003},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1845 pages = {179--214},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1846 abstract = {We recently conducted a research project for a large North American automobile insurer. This study was the most exhaustive ever undertaken by this particular insurer and lasted over an entire year. We analyzed the discriminating power of each variable used for ratemaking. We analyzed the performance of several models within five broad categories: linear regressions, generalized linear models, decision trees, neural networks and support vector machines. In this paper, we present the main results of this study. We qualitatively compare models and show how neural networks can represent high-order nonlinear dependencies with a small number of parameters, each of which is estimated on a large proportion of the data, thus yielding low variance. We thoroughly explain the purpose of the nonlinear sigmoidal transforms which are at the very heart of neural networks' performances. The main numerical result is a statistically significant reduction in the out-of-sample mean-squared error using the neural network model and our ability to substantially reduce the median premium by charging more to the highest risks. This in turn can translate into substantial savings and financial benefits for an insurer. We hope this paper goes a long way towards convincing actuaries to include neural networks within their set of modeling tools for ratemaking.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1847 topics={Finance,Mining},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1848 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1849
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1850 @INPROCEEDINGS{eck+bertinmahieux+lamere+green:nips2007,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1851 author = {Eck, Douglas and Lamere, Paul and Bertin-Mahieux, Thierry and Green, Stephen},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1852 editor = {Platt, John and Kolen, J. and Singer, Yoram and Roweis, S.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1853 title = {Automatic Generation of Social Tags for Music Recommendation},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1854 year = {2008},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1855 crossref = {NIPS20-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1856 source = "OwnPublication"
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1857 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1858
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1859 @INPROCEEDINGS{eck+bertinmahieux+lamere:ismir2007,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1860 author = {Eck, Douglas and Bertin-Mahieux, Thierry and Lamere, Paul},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1861 title = {Autotagging music using supervised machine learning},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1862 booktitle = {{Proceedings of the 8th International Conference on Music Information Retrieval ({ISMIR} 2007)}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1863 year = {2007},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1864 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1865 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1866
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1867 @INPROCEEDINGS{eck+casagrande:ismir2005,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1868 author = {Eck, Douglas and Casagrande, Norman},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1869 title = {Finding Meter in Music Using an Autocorrelation Phase Matrix and Shannon Entropy},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1870 booktitle = {{Proceedings of the 6th International Conference on Music Information Retrieval ({ISMIR} 2005)}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1871 year = {2005},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1872 pages = {504--509},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1873 url = {http://www.iro.umontreal.ca/~eckdoug/papers/2005_ismir.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1874 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1875 sourcetype={Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1876 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1877
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1878 @INCOLLECTION{eck+gasser+port:2000,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1879 author = {Eck, Douglas and Gasser, M. and Port, Robert},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1880 editor = {Desain, P. and Windsor, L.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1881 title = {Dynamics and Embodiment in Beat Induction},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1882 booktitle = {{Rhythm Perception and Production}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1883 year = {2000},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1884 pages = {157--170},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1885 publisher = {Swets and Zeitlinger},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1886 url = {http://www.iro.umontreal.ca/~eckdoug/papers/2000_rppw.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1887 abstract = {We provide an argument for using dynamical systems theory in the domain of beat induction. We motivate the study of beat induction and to relate beat induction to the more general study of human rhythm cognition. In doing so we compare a dynamical, embodied approach to a symbolic (traditional AI) one, paying particular attention to how the modeling approach brings with it tacit assumptions about what is being modeled. Please note that this is a philosophy paper about research that was, at the time of writing, very much in progress.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1888 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1889 sourcetype={Chapter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1890 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1891
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1892 @INPROCEEDINGS{eck+gasser:1996,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1893 author = {Eck, Douglas and Gasser, M.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1894 editor = {},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1895 title = {Perception of Simple Rhythmic Patterns in a Network of Oscillators},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1896 booktitle = {{The Proceedings of the Eighteenth Annual Conference of the Cognitive Science Society}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1897 year = {1996},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1898 publisher = {Lawrence Erlbaum Associates},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1899 abstract = {This paper is concerned with the complex capacity to recognize and reproduce rhythmic patterns. While this capacity has not been well investigated, in broad qualitative terms it is clear that people can learn to identify and produce recurring patterns defined in terms of sequences of beats of varying intensity and rests: the rhythms behind waltzes, reels, sambas, etc. Our short term goal is a model which is "hard-wired" with knowledge of a set of such patterns. Presented with a portion of one of the patterns or a label for a pattern, the model should reproduce the pattern and continue to do so when the input is turned off. Our long-term goal is a model which can learn to adjust the connection strengths which implement particular patterns as it is exposed to input patterns.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1900 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1901 sourcetype={Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1902 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1903
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1904 @TECHREPORT{eck+graves+schmidhuber:tr-speech2003,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1905 author = {Eck, Douglas and Graves, A. and Schmidhuber, Juergen},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1906 title = {A New Approach to Continuous Speech Recognition Using {LSTM} Recurrent Neural Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1907 number = {IDSIA-14-03},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1908 year = {2003},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1909 institution = {IDSIA},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1910 abstract = {This paper presents an algorithm for continuous speech recognition built from two Long Short-Term Memory ({LSTM}) recurrent neural networks. A first {LSTM} network performs frame-level phone probability estimation. A second network maps these phone predictions onto words. In contrast to {HMM}s, this allows greater exploitation of long-timescale correlations. Simulation results are presented for a hand-segmented subset of the "Numbers-95" database. These results include isolated phone prediction, continuous frame-level phone prediction and continuous word prediction. We conclude that despite its early stage of development, our new model is already competitive with existing approaches on certain aspects of speech recognition and promising on others, warranting further research.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1911 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1912 sourcetype={TechReport},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1913 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1914
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1915 @TECHREPORT{eck+lapalme:2008,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1916 author = {Eck, Douglas and Lapalme, J.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1917 title = {Learning Musical Structure Directly from Sequences of Music},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1918 number = {1300},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1919 year = {2008},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1920 institution = {Universit{\'{e}} de Montr{\'{e}}al DIRO},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1921 url = {http://www.iro.umontreal.ca/~eckdoug/papers/tr1300.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1922 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1923 sourcetype={TechReport},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1924 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1925
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1926 @INPROCEEDINGS{eck+schmidhuber:icann2002,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1927 author = {Eck, Douglas and Schmidhuber, Juergen},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1928 editor = {Dorronsoro, J.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1929 title = {Learning The Long-Term Structure of the Blues},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1930 booktitle = {{Artificial Neural Networks -- ICANN 2002 (Proceedings)}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1931 volume = {},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1932 year = {2002},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1933 pages = {284--289},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1934 publisher = {Springer},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1935 url = {http://www.iro.umontreal.ca/~eckdoug/papers/2002_icannMusic.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1936 abstract = {In general music composed by recurrent neural networks ({RNN}s) suffers from a lack of global structure. Though networks can learn note-by-note transition probabilities and even reproduce phrases, they have been unable to learn an entire musical form and use that knowledge to guide composition. In this study, we describe model details and present experimental results showing that {LSTM} successfully learns a form of blues music and is able to compose novel (and some listeners believe pleasing) melodies in that style. Remarkably, once the network has found the relevant structure it does not drift from it: {LSTM} is able to play the blues with good timing and proper structure as long as one is willing to listen.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1937 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1938 sourcetype={Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1939 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1940
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1941 @INPROCEEDINGS{eck+schmidhuber:ieee2002,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1942 author = {Eck, Douglas and Schmidhuber, Juergen},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1943 editor = {Bourlard, H.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1944 title = {Finding Temporal Structure in Music: Blues Improvisation with {LSTM} Recurrent Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1945 booktitle = {Neural Networks for Signal Processing XII, Proceedings of the 2002 IEEE Workshop},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1946 year = {2002},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1947 pages = {747--756},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1948 publisher = {IEEE},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1949 url = {http://www.iro.umontreal.ca/~eckdoug/papers/2002_ieee.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1950 abstract = {Few types of signal streams are as ubiquitous as music. Here we consider the problem of extracting essential ingredients of music signals, such as well-defined global temporal structure in the form of nested periodicities (or {\em meter}). Can we construct an adaptive signal processing device that learns by example how to generate new instances of a given musical style? Because recurrent neural networks can in principle learn the temporal structure of a signal, they are good candidates for such a task. Unfortunately, music composed by standard recurrent neural networks ({RNN}s) often lacks global coherence. The reason for this failure seems to be that {RNN}s cannot keep track of temporally distant events that indicate global music structure. Long Short-Term Memory ({LSTM}) has succeeded in similar domains where other {RNN}s have failed, such as timing \& counting and learning of context sensitive languages. In the current study we show that {LSTM} is also a good mechanism for learning to compose music. We present experimental results showing that {LSTM} successfully learns a form of blues music and is able to compose novel (and we believe pleasing) melodies in that style. Remarkably, once the network has found the relevant structure it does not drift from it: {LSTM} is able to play the blues with good timing and proper structure as long as one is willing to listen.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1951 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1952 sourcetype={Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1953 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1954
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1955 @ARTICLE{eck+scott:2005,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1956 author = {Eck, Douglas and Scott, S. K.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1957 title = {Editorial: New Research in Rhythm Perception and Production},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1958 journal = {Music Perception},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1959 volume = {22},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1960 number = {3},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1961 year = {2005},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1962 pages = {371-388},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1963 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1964 sourcetype={Other},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1965 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1966
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1967 @MISC{eck+scott:editor2005,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1968 author = {Eck, Douglas and Scott, S. K.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1969 title = {Music Perception},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1970 year = {2005},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1971 note = {Guest Editor, Special Issue on Rhythm Perception and Production, 22(3)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1972 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1973 sourcetype={Other},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1974 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1975
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1976 @INPROCEEDINGS{eck:1999,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1977 author = {Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1978 editor = {},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1979 title = {Learning Simple Metrical Preferences in a Network of {F}itzhugh-{N}agumo Oscillators},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1980 booktitle = {{The Proceedings of the Twenty-First Annual Conference of the Cognitive Science Society}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1981 year = {1999},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1982 publisher = {Lawrence Erlbaum Associates},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1983 abstract = {Hebbian learning is used to train a network of oscillators to prefer periodic signals of pulses over aperiodic signals. Target signals consisted of metronome-like voltage pulses with varying amounts of inter-onset noise injected. (with 0\% noise yielding a periodic signal and more noise yielding more and more aperiodic signals.) The oscillators---piecewise-linear approximations (Abbott, 1990) to Fitzhugh-Nagumo oscillators---are trained using mean phase coherence as an objective function. Before training a network is shown to readily synchronize with signals having wide range of noise. After training on a series of noise-free signals, a network is shown to only synchronize with signals having little or no noise. This represents a bias towards periodicity and is explained by strong positive coupling connections between oscillators having harmonically-related periods.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1984 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1985 sourcetype={Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1986 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1987
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1988 @UNPUBLISHED{eck:bramsworkshop2004,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1989 author = {Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1990 title = {Challenges for Machine Learning in the Domain of Music},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1991 year = {2004},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1992 note = {BRAMS Workshop on Brain and Music, Montreal Neurological Institute},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1993 abstract = {Slides and musical examples available on request.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1994 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1995 sourcetype={Workshop},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1996 optkey={""},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1997 optmonth={""},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1998 optannote={""},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	1999 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2000
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2001 @PHDTHESIS{eck:diss,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2002 author = {Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2003 title = {{Meter Through Synchrony: Processing Rhythmical Patterns with Relaxation Oscillators}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2004 year = {2000},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2005 school = {Indiana University, Bloomington, IN, www.idsia.ch/\-\~{}doug/\-publications.html},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2006 abstract = {This dissertation uses a network of relaxation oscillators to beat along with temporal signals. Relaxation oscillators exhibit interspersed slow-fast movement and model a wide array of biological oscillations. The model is built up gradually: first a single relaxation oscillator is exposed to rhythms and shown to be good at finding downbeats in them. Then large networks of oscillators are mutually coupled in an exploration of their internal synchronization behavior. It is demonstrated that appropriate weights on coupling connections cause a network to form multiple pools of oscillators having stable phase relationships. This is a promising first step towards networks that can recreate a rhythmical pattern from memory. In the full model, a coupled network of relaxation oscillators is exposed to rhythmical patterns. It is shown that the network finds downbeats in patterns while continuing to exhibit good internal stability. A novel non-dynamical model of downbeat induction called the Normalized Positive (NP) clock model is proposed, analyzed, and used to generate comparison predictions for the oscillator model. The oscillator model compares favorably to other dynamical approaches to beat induction such as adaptive oscillators. However, the relaxation oscillator model takes advantage of intrinsic synchronization stability to allow the creation of large coupled networks. This research lays the groundwork for a long-term research goal, a robotic arm that responds to rhythmical signals by tapping along. It also opens the door to future work in connectionist learning of long rhythmical patterns.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2007 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2008 sourcetype={Thesis},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2009 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2010
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2011 @INPROCEEDINGS{eck:icann2001,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2012 author = {Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2013 editor = {Dorffner, Georg},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2014 title = {A Network of Relaxation Oscillators that Finds Downbeats in Rhythms},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2015 booktitle = {{Artificial Neural Networks -- ICANN 2001 (Proceedings)}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2016 volume = {},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2017 year = {2001},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2018 pages = {1239--1247},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2019 publisher = {Springer},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2020 url = {http://www.iro.umontreal.ca/~eckdoug/papers/2001_icann.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2021 abstract = {A network of relaxation oscillators is used to find downbeats in rhythmical patterns. In this study, a novel model is described in detail. Its behavior is tested by exposing it to patterns having various levels of rhythmic complexity. We analyze the performance of the model and relate its success to previous work dealing with fast synchrony in coupled oscillators.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2022 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2023 sourcetype={Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2024 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2025
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2026 @INPROCEEDINGS{eck:icassp2007,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2027 author = {Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2028 editor = {},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2029 title = {Beat Tracking Using an Autocorrelation Phase Matrix},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2030 booktitle = {{Proceedings of the 2007 International Conference on Acoustics, Speech and Signal Processing (ICASSP)}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2031 year = {2007},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2032 pages = {1313--1316},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2033 publisher = {IEEE Signal Processing Society},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2034 url = {http://www.iro.umontreal.ca/~eckdoug/papers/2007_icassp.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2035 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2036 sourcetype={Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2037 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2038
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2039 @INPROCEEDINGS{eck:icmpc2004,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2040 author = {Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2041 editor = {Lipscomb, S. D. and Ashley, R. and Gjerdingen, R. O. and Webster, P.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2042 title = {A Machine-Learning Approach to Musical Sequence Induction That Uses Autocorrelation to Bridge Long Timelags},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2043 booktitle = {{The Proceedings of the Eighth International Conference on Music Perception and Cognition ({ICMPC}8)}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2044 year = {2004},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2045 pages = {542-543},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2046 publisher = {Causal Productions},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2047 abstract = {One major challenge in using statistical sequence learning methods in the domain of music lies in bridging the long timelags that separate important musical events. Consider, for example, the chord changes that convey the basic structure of a pop song. A sequence learner that cannot predict chord changes will almost certainly not be able to generate new examples in a musical style or to categorize songs by style. Yet, it is surprisingly difficult for a sequence learner to bridge the long timelags necessary to identify when a chord change will occur and what its new value will be. This is the case because chord changes can be separated by dozens or hundreds of intervening notes. One could solve this problem by treating chords as being special (as did Mozer, NIPS 1991). But this is impractical---it requires chords to be labeled specially in the dataset, limiting the applicability of the model to non-labeled examples---and furthermore does not address the general issue of nested temporal structure in music. I will briefly describe this temporal structure (known commonly as "meter") and present a model that uses to its advantage an assumption that sequences are metrical. The model consists of an autocorrelation-based filtration that estimates online the most likely metrical tree (i.e. the frequency and phase of beat, measure, phrase &etc.) and uses that to generate a series of sequences varying at different rates. These sequences correspond to each level in the hierarchy. Multiple learners can be used to treat each series separately and their predictions can be combined to perform composition and categorization. I will present preliminary results that demonstrate the usefulness of this approach. Time permitting I will also compare the model to alternate approaches.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2048 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2049 sourcetype={Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2050 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2051
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2052 @INPROCEEDINGS{eck:icmpc2006,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2053 author = {Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2054 editor = {Baroni, M. and Addessi, A. R. and Caterina, R. and Costa, M.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2055 title = {Beat Induction Using an Autocorrelation Phase Matrix},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2056 booktitle = {The Proceedings of the 9th International Conference on Music Perception and Cognition ({ICMPC9})},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2057 year = {2006},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2058 pages = {931-932},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2059 publisher = {Causal Productions},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2060 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2061 sourcetype={Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2062 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2063
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2064 @UNPUBLISHED{eck:irisworkshop2004,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2065 author = {Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2066 title = {Using Autocorrelation to Bridge Long Timelags when Learning Sequences of Music},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2067 year = {2004},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2068 note = {IRIS 2004 Machine Learning Workshop, Ottawa, Canada},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2069 abstract = {Slides and musical examples available on request.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2070 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2071 sourcetype={Workshop},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2072 optkey={""},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2073 optmonth={""},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2074 optannote={""},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2075 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2076
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2077 @ARTICLE{eck:jnmr2001,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2078 author = {Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2079 title = {A Positive-Evidence Model for Rhythmical Beat Induction},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2080 journal = {Journal of New Music Research},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2081 volume = {30},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2082 number = {2},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2083 year = {2001},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2084 pages = {187--200},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2085 abstract = {The Normalized Positive (NPOS) model is a rule-based model that predicts downbeat location and pattern complexity in rhythmical patterns. Though derived from several existing models, the NPOS model is particularly effective at making correct predictions while at the same time having low complexity. In this paper, the details of the model are explored and a comparison is made to existing models. Several datasets are used to examine the complexity predictions of the model. Special attention is paid to the model's ability to account for the effects of musical experience on beat induction.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2086 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2087 sourcetype={Journal},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2088 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2089
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2090 @UNPUBLISHED{eck:mipsworkshop2004,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2091 author = {Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2092 title = {Bridging Long Timelags in Music},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2093 year = {2004},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2094 note = {NIPS 2004 Workshop on Music and Machine Learning (MIPS), Whistler, British Columbia},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2095 abstract = {Slides and musical examples available on request.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2096 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2097 sourcetype={Workshop},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2098 optkey={""},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2099 optmonth={""},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2100 optannote={""},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2101 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2102
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2103 @ARTICLE{eck:mp2006,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2104 author = {Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2105 title = {Finding Long-Timescale Musical Structure with an Autocorrelation Phase Matrix},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2106 journal = {Music Perception},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2107 volume = {24},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2108 number = {2},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2109 year = {2006},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2110 pages = {167--176},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2111 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2112 sourcetype={Journal},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2113 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2114
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2115 @UNPUBLISHED{eck:nipsworkshop2003,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2116 author = {Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2117 title = {Time-warped hierarchical structure in music and speech: A sequence prediction challenge},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2118 year = {2003},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2119 note = {NIPS 2003 Workshop on Recurrent Neural Networks, Whistler, British Columbia},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2120 abstract = {Slides and musical examples available on request.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2121 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2122 sourcetype={Workshop},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2123 optkey={""},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2124 optmonth={""},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2125 optannote={""},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2126 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2127
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2128 @UNPUBLISHED{eck:nipsworkshop2006,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2129 author = {Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2130 title = {Generating music sequences with an echo state network},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2131 year = {2006},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2132 note = {NIPS 2006 Workshop on Echo State Networks and Liquid State Machines},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2133 abstract = {Slides and musical examples available on request.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2134 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2135 sourcetype={Workshop},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2136 optkey={""},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2137 optmonth={""},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2138 optannote={""},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2139 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2140
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2141 @UNPUBLISHED{eck:nipsworkshop2007,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2142 author = {Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2143 title = {Measuring and modeling musical expression},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2144 year = {2007},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2145 note = {NIPS 2007 Workshop on Music, Brain and Cognition},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2146 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2147 sourcetype={Workshop},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2148 optkey={""},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2149 optmonth={""},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2150 optannote={""},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2151 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2152
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2153 @ARTICLE{eck:psyres2002,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2154 author = {Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2155 title = {Finding Downbeats with a Relaxation Oscillator},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2156 journal = {Psychol. Research},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2157 volume = {66},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2158 number = {1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2159 year = {2002},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2160 pages = {18--25},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2161 abstract = {A relaxation oscillator model of neural spiking dynamics is applied to the task of finding downbeats in rhythmical patterns. The importance of downbeat discovery or {\em beat induction} is discussed, and the relaxation oscillator model is compared to other oscillator models. In a set of computer simulations the model is tested on 35 rhythmical patterns from Povel \& Essens (1985). The model performs well, making good predictions in 34 of 35 cases. In an analysis we identify some shortcomings of the model and relate model behavior to dynamical properties of relaxation oscillators.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2162 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2163 sourcetype={Journal},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2164 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2165
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2166 @UNPUBLISHED{eck:rppw2005,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2167 author = {Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2168 title = {Meter and Autocorrelation},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2169 year = {2005},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2170 note = {{10th Rhythm Perception and Production Workshop (RPPW), Alden Biesen, Belgium}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2171 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2172 sourcetype={Workshop},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2173 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2174
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2175 @TECHREPORT{eck:tr-music2002,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2176 author = {Eck, Douglas and Schmidhuber, Juergen},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2177 title = {A First Look at Music Composition using {LSTM} Recurrent Neural Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2178 number = {IDSIA-07-02},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2179 year = {2002},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2180 institution = {IDSIA},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2181 abstract = {In general music composed by recurrent neural networks ({RNN}s) suffers from a lack of global structure. Though networks can learn note-by-note transition probabilities and even reproduce phrases, attempts at learning an entire musical form and using that knowledge to guide composition have been unsuccessful. The reason for this failure seems to be that {RNN}s cannot keep track of temporally distant events that indicate global music structure. Long Short-Term Memory ({LSTM}) has succeeded in similar domains where other {RNN}s have failed, such as timing \& counting and CSL learning. In the current study I show that {LSTM} is also a good mechanism for learning to compose music. I compare this approach to previous attempts, with particular focus on issues of data representation. I present experimental results showing that {LSTM} successfully learns a form of blues music and is able to compose novel (and I believe pleasing) melodies in that style. Remarkably, once the network has found the relevant structure it does not drift from it: {LSTM} is able to play the blues with good timing and proper structure as long as one is willing to listen. {\em Note: This is a more complete version of the 2002 ICANN submission Learning the Long-Term Structure of the Blues.}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2182 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2183 sourcetype={TechReport},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2184 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2185
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2186 @TECHREPORT{eck:tr-npos2000,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2187 author = {Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2188 title = {A Positive-Evidence Model for Classifying Rhythmical Patterns},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2189 number = {IDSIA-09-00},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2190 year = {2000},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2191 institution = {IDSIA},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2192 abstract = {The Normalized Positive (NPOS) model is a novel matching model that predicts downbeat location and pattern complexity in rhythmical patterns. Though similar models report success, the NPOS model is particularly effective at making these predictions while at the same time being theoretically and mathematically simple. In this paper, the details of the model are explored and a comparison is made to existing models. Several datasets are used to examine the complexity predictions of the model. Special attention is paid to the model's ability to account for the effects of musical experience on rhythm perception.\\ {\em Note: See the 2001 Journal of New Music Research paper "A Positive-Evidence Model for Rhythmical Beat Induction" for a newer version of this paper.}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2193 ps={ftp://ftp.idsia.ch/pub/techrep/IDSIA-09-00.ps.gz},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2194 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2195 sourcetype={TechReport},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2196 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2197
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2198 @TECHREPORT{eck:tr-oscnet2001,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2199 author = {Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2200 title = {A Network of Relaxation Oscillators that Finds Downbeats in Rhythms},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2201 number = {IDSIA-06-01},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2202 year = {2001},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2203 institution = {IDSIA},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2204 abstract = {A network of relaxation oscillators is used to find downbeats in rhythmical patterns. In this study, a novel model is described in detail. Its behavior is tested by exposing it to patterns having various levels of rhythmic complexity. We analyze the performance of the model and relate its success to previous work dealing with fast synchrony in coupled oscillators. \\ {\em Note: See the 2001 ICANN conference proceeding by the same title for a newer version of this paper.}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2205 ps={ftp://ftp.idsia.ch/pub/techrep/IDSIA-06-01.ps.gz},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2206 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2207 sourcetype={TechReport},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2208 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2209
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2210 @TECHREPORT{eck:tr-tracking2000,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2211 author = {Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2212 title = {Tracking Rhythms with a Relaxation Oscillator},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2213 number = {IDSIA-10-00},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2214 year = {2000},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2215 institution = {IDSIA},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2216 abstract = {A number of biological and mechanical processes are typified by a continued slow accrual and fast release of energy. A nonlinear oscillator exhibiting this slow-fast behavior is called a relaxation oscillator and is used to model, for example, human heartbeat pacemaking and neural action potential. Similar limit cycle oscillators are used to model a wider range of behaviors including predator-prey relationships and synchrony in animal populations such as fireflies. Though nonlinear limit-cycle oscillators have been successfully applied to beat induction, relaxation oscillators have received less attention. In this work we offer a novel and effective relaxation oscillator model of beat induction. We outline the model in detail and provide a perturbation analysis of its response to external stimuli. In a series of simulations we expose the model to patterns from Experiment 1 of Povel \& Essens (1985). We then examine the beat assignments of the model. Although the overall performance of the model is very good, there are shortcomings. We believe that a network of mutually-coupled oscillators will address many of these shortcomings, and we suggest an appropriate course for future research.\\ {\em Note: See the 2001 {\em Psychological Research} article "Finding Downbeats with a Relaxation Oscillator" for a revised but less detailed version of this paper.}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2217 ps={ftp://ftp.idsia.ch/pub/techrep/IDSIA-10-00.ps.gz},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2218 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2219 sourcetype={TechReport},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2220 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2221
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2222 @TECHREPORT{eck:tr-tracking2002,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2223 author = {Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2224 title = {Real-Time Musical Beat Induction with Spiking Neural Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2225 number = {IDSIA-22-02},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2226 year = {2002},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2227 institution = {IDSIA},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2228 abstract = {Beat induction is best described by analogy to the activities of hand clapping or foot tapping, and involves finding important metrical components in an auditory signal, usually music. Though beat induction is intuitively easy to understand it is difficult to define and still more difficult to perform automatically. We will present a model of beat induction that uses a spiking neural network as the underlying synchronization mechanism. This approach has some advantages over existing methods; it runs online, responds at many levels in the metrical hierarchy, and produces good results on performed music (Beatles piano performances encoded as MIDI). In this paper the model is described in some detail and simulation results are discussed.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2229 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2230 sourcetype={TechReport},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2231 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2232
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2233 @UNPUBLISHED{eck:verita2002,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2234 author = {Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2235 title = {Real Time Beat Induction with Spiking Neurons},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2236 year = {2002},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2237 note = {{Music, Motor Control and the Mind: Symposium at Monte Verita, May}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2238 abstract = {Beat induction is best described by analogy to the activites of hand clapping or foot tapping, and involves finding important metrical components in an auditory signal, usually music. Though beat induction is intuitively easy to understand it is difficult to define and still more difficult to model. I will discuss an approach to beat induction that uses a network of spiking neurons to synchronize with periodic components in a signal at many timescales. Through a competitive process, groups of oscillators embodying a particular metrical interpretation (e.g. \"4/4\") are selected from the network and used to track the pattern. I will compare this model to other approaches including a traditional symbolic AI system (Dixon 2001), and one based on Bayesian statistics (Cemgil et al, 2001). Finally I will present performance results of the network on a set of MIDI-recorded piano performances of Beatles songs collected by the Music, Mind, Machine Group, NICI, University of Nijmegen (see Cemgil et al, 2001 for more details or http://www.nici.kun.nl/mmm).},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2239 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2240 sourcetype={Workshop},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2241 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2242
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2243 @INPROCEEDINGS{ElHihi+Bengio-nips8,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2244 author = {El Hihi, Salah and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2245 title = {Hierarchical Recurrent Neural Networks for Long-Term Dependencies},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2246 year = {1996},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2247 crossref = {NIPS8-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2248 abstract = {We have already shown that extracting lone-term dependencies from sequential data is difficult, both for deterministic dynamical systems such as recurrent networks, and probabilistic models such as hidden {Markov} models ({HMM}s) or input/output hidden {Markov} models ({IOHMM}s). In practice, to avoid this problem, researchers have used domain specific a-priori knowledge to give meaning to the hidden or state variables representing past context. In this paper we propose to use a more general type of a-priori knowledge, namely that the temporal dependencies are structured hierarchically. This implies that long-term dependencies are represented by variables with a long time scale. This principle is applied to a recurrent network which includes delays and multiple time scales. Experiments confirm the advantages of such structures. A similar approach is proposed for {HMM}s and {IOHMM}s.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2249 topics={LongTerm},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2250 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2251
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2252 @ARTICLE{Erhan+al-2010,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2253 author = {Erhan, Dumitru and Bengio, Yoshua and Courville, Aaron and Manzagol, Pierre-Antoine and Vincent, Pascal and Bengio, Samy},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2254 title = {Why Does Unsupervised Pre-training Help Deep Learning?},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2255 volume = {11},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2256 year = {2010},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2257 pages = {625--660},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2258 journal = {Journal of Machine Learning Research},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2259 abstract = {Much recent research has been devoted to learning algorithms for deep architectures such as Deep Belief Networks and stacks of auto-encoder variants, with impressive results obtained in several areas, mostly on vision and language datasets. The best results obtained on supervised learning tasks involve an unsupervised learning component, usually in an unsupervised pre-training phase. Even though these new algorithms have enabled training deep models, many questions remain as to the nature of this difficult learning problem. The main question investigated here is the following: why does unsupervised pre-training work and why does it work so well? Answering these questions is important if learning in deep architectures is to be further improved. We propose several explanatory hypotheses and test them through extensive simulations. We empirically show the influence of pre-training with respect to architecture depth, model capacity, and number of training examples. The experiments confirm and clarify the advantage of unsupervised pre-training. The results suggest that unsupervised pre-training guides the learning towards basins of attraction of minima that are better in terms of the underlying data distribution; the evidence from these results supports a regularization explanation for the effect of pre-training.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2260 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2261
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2262 @INPROCEEDINGS{Erhan-aistats-2010,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2263 author = {Erhan, Dumitru and Courville, Aaron and Bengio, Yoshua and Vincent, Pascal},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2264 title = {Why Does Unsupervised Pre-training Help Deep Learning?},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2265 booktitle = {Proceedings of AISTATS 2010},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2266 volume = {9},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2267 year = {2010},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2268 pages = {201-208},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2269 abstract = {Much recent research has been devoted to learning algorithms for deep architectures such as Deep Belief Networks and stacks of auto-encoder variants with impressive results being obtained in several areas, mostly on vision and language datasets. The best results obtained on supervised learning tasks often involve an unsupervised learning component, usually in an unsupervised pre-training phase. The main question investigated here is the following: why does unsupervised pre-training work so well? Through extensive experimentation, we explore several possible explanations discussed in the literature including its action as a regularizer (Erhan et al. 2009) and as an aid to optimization (Bengio et al. 2007). Our results build on the work of Erhan et al. 2009, showing that unsupervised pre-training appears to play predominantly a regularization role in subsequent supervised training. However our results in an online setting, with a virtually unlimited data stream, point to a somewhat more nuanced interpretation of the roles of optimization and regularization in the unsupervised pre-training effect.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2270 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2271
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2272 @MASTERSTHESIS{Erhan-MSc,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2273 author = {Erhan, Dumitru},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2274 keywords = {Apprentisage multit{\^{a}}che, Filtrage collaboratif, M{\'{e}}thodes {\`{a}} noyaux, QSAR, R{\'{e}}seaux de neurones},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2275 title = {Collaborative filtering techniques for drug discovery},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2276 year = {2006},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2277 school = {Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2278 abstract = {Cette th{\`{e}}se examine le probl{\`{e}}me d'apprendre plusieurs t{\^{a}}ches simultan{\'{e}}ment,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2279 afin de transf{\'{e}}rer les connaissances apprises {\`{a}} une nouvelle t{\^{a}}che. Si
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2280 on suppose que les t{\^{a}}ches partagent une repr{\'{e}}sentation et qu'il est possible de
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2281 d{\'{e}}couvrir cette repr{\'{e}}sentation efficacement, cela peut nous servir {\`{a}} construire un
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2282 meilleur mod{\`{e}}le de la nouvelle t{\^{a}}che. Il existe plusieurs variantes de
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2283 cette m{\'{e}}thode: transfert inductif, apprentisage multit{\^{a}}che, filtrage
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2284 collaboratif etc. Nous avons {\'{e}}valu{\'{e}} plusieurs algorithmes d'apprentisage
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2285 supervis{\'{e}} pour d{\'{e}}couvrir des repr{\'{e}}sentations partag{\'{e}}es parmi les
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2286 t{\^{a}}ches d{\'{e}}finies dans un probl{\`{e}}me de chimie computationelle. Nous avons
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2287 formul{\'{e}} le probl{\`{e}}me dans un cadre d'apprentisage automatique,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2288 fait l'analogie avec les algorithmes standards de filtrage collaboratif et construit les
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2289 hypoth{\`{e}}ses g{\'{e}}n{\'{e}}rales qui devraient {\^{e}}tre test{\'{e}}es pour valider l'utilitisation des
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2290 algorithmes multit{\^{a}}che. Nous avons aussi {\'{e}}valu{\'{e}} la performance des algorithmes
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2291 d'apprentisage utilis{\'{e}}s et d{\'{e}}montrons qu'il est, en effet, possible de trouver une
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2292 repr{\'{e}}sentation partag{\'{e}}e pour le probl{\`{e}}me consider{\'{e}}. Du point de vue
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2293 th{\'{e}}orique, notre apport est une modification d'un algorithme
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2294 standard---les machines {\`{a}} vecteurs de support--qui produit des r{\'{e}}sultats
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2295 comparables aux meilleurs algorithmes disponsibles et qui utilise {\`{a}} fond les
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2296 concepts de l'apprentisage multit{\^{a}}che. Du point de vue pratique, notre
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2297 apport est l'utilisation de notre algorithme par les compagnies
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2298 pharmaceutiques dans leur d{\'{e}}couverte de nouveaux m{\'{e}}dicaments.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2299 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2300
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2301 @INPROCEEDINGS{Erhan2009,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2302 author = {Erhan, Dumitru and Manzagol, Pierre-Antoine and Bengio, Yoshua and Bengio, Samy and Vincent, Pascal},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2303 keywords = {Deep Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2304 title = {The Difficulty of Training Deep Architectures and the effect of Unsupervised Pre-Training},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2305 year = {2009},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2306 pages = {153--160},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2307 crossref = {xAISTATS2009-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2308 abstract = {Whereas theoretical work suggests that deep architectures might be more efﬁcient at representing highly-varying functions, training deep architectures was unsuccessful until the recent advent of algorithms based on unsupervised pretraining. Even though these new algorithms have enabled training deep models, many questions remain as to the nature of this difﬁcult learning problem. Answering these questions is important if learning in deep architectures is to be further improved. We attempt to shed some light on these questions through extensive simulations. The experiments conﬁrm and clarify the advantage of unsupervised pre-training. They demonstrate the robustness of the training procedure with respect to the random initialization, the positive effect of pre-training in terms of optimization and its role as a regularizer. We empirically show the inﬂuence of pre-training with respect to architecture depth, model capacity, and number of training examples.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2309 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2310
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2311 @ARTICLE{gasser+eck+port:1999,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2312 author = {Gasser, M. and Eck, Douglas and Port, Robert},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2313 title = {Meter as Mechanism: A Neural Network Model that Learns Metrical patterns},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2314 journal = {Connection Science},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2315 volume = {11},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2316 number = {2},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2317 year = {1999},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2318 pages = {187--216},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2319 abstract = {One kind of prosodic structure that apparently underlies both music and some examples of speech production is meter. Yet detailed measurements of the timing of both music and speech show that the nested periodicities that define metrical structure can be quite noisy in time. What kind of system could produce or perceive such variable metrical timing patterns? And what would it take to be able to store and reproduce particular metrical patterns from long-term memory? We have developed a network of coupled oscillators that both produces and perceives patterns of pulses that conform to particular meters. In addition, beginning with an initial state with no biases, it can learn to prefer the particular meter that it has been previously exposed to.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2320 own={Have},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2321 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2322 sourcetype={Journal},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2323 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2324
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2325 @TECHREPORT{gasser+eck+port:tr-1996,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2326 author = {Gasser, M. and Eck, Douglas and Port, Robert},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2327 title = {Meter as Mechanism A Neural Network that Learns Metrical Patterns},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2328 number = {180},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2329 year = {1996},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2330 institution = {Indiana University Cognitive Science Program},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2331 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2332 sourcetype={TechReport},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2333 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2334
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2335 @INPROCEEDINGS{gasser+eck:1996,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2336 author = {Gasser, M. and Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2337 editor = {},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2338 title = {Representing Rhythmic Patterns in a Network of Oscillators},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2339 booktitle = {{The Proceedings of the International Conference on Music Perception and Cognition}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2340 number = {4},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2341 year = {1996},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2342 pages = {361--366},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2343 publisher = {Lawrence Erlbaum Associates},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2344 url = {http://www.iro.umontreal.ca/~eckdoug/papers/1996_gasser_icmpc.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2345 abstract = {This paper describes an evolving computational model of the perception and pro-duction of simple rhythmic patterns. The model consists of a network of oscillators of different resting frequencies which couple with input patterns and with each other. Os-cillators whose frequencies match periodicities in the input tend to become activated. Metrical structure is represented explicitly in the network in the form of clusters of os-cillators whose frequencies and phase angles are constrained to maintain the harmonic relationships that characterize meter. Rests in rhythmic patterns are represented by ex-plicit rest oscillators in the network, which become activated when an expected beat in the pattern fails to appear. The model makes predictions about the relative difficulty of patterns and the effect of deviations from periodicity in the input.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2346 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2347 sourcetype={Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2348 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2349
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2350 @INPROCEEDINGS{gers+eck+schmidhuber:icann2001,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2351 author = {Gers, F. A. and Eck, Douglas and Schmidhuber, Juergen},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2352 editor = {Dorffner, Georg},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2353 title = {Applying {LSTM} to Time Series Predictable Through Time-Window Approaches},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2354 booktitle = {{Artificial Neural Networks -- ICANN 2001 (Proceedings)}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2355 year = {2001},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2356 pages = {669--676},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2357 publisher = {Springer},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2358 url = {http://www.iro.umontreal.ca/~eckdoug/papers/2001_gers_icann.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2359 abstract = {Long Short-Term Memory ({LSTM}) is able to solve many time series tasks unsolvable by feed-forward networks using fixed size time windows. Here we find that {LSTM}'s superiority does {\em not} carry over to certain simpler time series tasks solvable by time window approaches: the Mackey-Glass series and the Santa Fe {FIR} laser emission series (Set A). This suggests t use {LSTM} only when simpler traditional approaches fail.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2360 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2361 sourcetype={Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2362 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2363
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2364 @TECHREPORT{gers+eck+schmidhuber:tr-2000,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2365 author = {Gers, F. A. and Eck, Douglas and Schmidhuber, Juergen},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2366 title = {Applying {LSTM} to Time Series Predictable Through Time-Window Approaches},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2367 number = {IDSIA-22-00},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2368 year = {2000},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2369 institution = {IDSIA},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2370 abstract = {Long Short-Term Memory ({LSTM}) is able to solve many time series tasks unsolvable by feed-forward networks using fixed size time windows. Here we find that {LSTM}'s superiority does {\em not} carry over to certain simpler time series tasks solvable by time window approaches: the Mackey-Glass series and the Santa Fe {FIR} laser emission series (Set A). This suggests t use {LSTM} only when simpler traditional approaches fail.\\ {\em Note: See the 2001 ICANN conference proceeding by the same title for a newer version of this paper.}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2371 ps={ftp://ftp.idsia.ch/pub/techrep/IDSIA-22-00.ps.gz},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2372 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2373 sourcetype={TechReport},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2374 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2375
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2376 @INPROCEEDINGS{gers+perez+eck+schmidhuber:esann2002,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2377 author = {Gers, F. A. and Perez-Ortiz, J. A. and Eck, Douglas and Schmidhuber, Juergen},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2378 title = {{DEKF-LSTM}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2379 booktitle = {Proceedings of the 10th European Symposium on Artificial Neural Networks, ESANN 2002},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2380 year = {2002},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2381 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2382 sourcetype={Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2383 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2384
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2385 @INPROCEEDINGS{gers+perez+eck+schmidhuber:icannA2002,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2386 author = {Gers, F. A. and Perez-Ortiz, J. A. and Eck, Douglas and Schmidhuber, Juergen},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2387 editor = {Dorronsoro, J.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2388 title = {Learning Context Sensitive Languages with {LSTM} Trained with {Kalman} Filters},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2389 booktitle = {{Artificial Neural Networks -- ICANN 2002 (Proceedings)}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2390 year = {2002},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2391 pages = {655--660},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2392 publisher = {Springer},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2393 abstract = {Unlike traditional recurrent neural networks, the Long Short-Term Memory ({LSTM}) model generalizes well when presented with training sequences derived from regular and also simple nonregular languages. Our novel combination of {LSTM} and the decoupled extended Kalman filter, however, learns even faster and generalizes even better, requiring only the 10 shortest exemplars n <= 10 of the context sensitive language a^nb^nc^n to deal correctly with values of n up to 1000 and more. Even when we consider the relatively high update complexity per timestep, in many cases the hybrid offers faster learning than {LSTM} by itself.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2394 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2395 sourcetype={Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2396 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2397
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2398 @PHDTHESIS{Ghosn-Phd-2003,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2399 author = {Ghosn, Joumana},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2400 title = {Apprentissage multi-t{\^{a}}ches et partage de connaissances},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2401 year = {2003},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2402 school = {Universit{\'{e}} de Montr{\'{e}}al}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2403 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2404
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2405 @INPROCEEDINGS{ghosn97,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2406 author = {Ghosn, Joumana and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2407 title = {Multi-Task Learning for Stock Selection},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2408 year = {1997},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2409 pages = {946--952},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2410 publisher = {MIT Press, Cambridge, MA},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2411 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/multitask-nips97.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2412 crossref = {NIPS9},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2413 abstract = {Artificial Neural Networks can be used to predict future returns of stocks in order to take financial decisions. Should one build a separate network for each stock or share the same network for all the stocks. In this paper we also explore other alternatives, in which some layers are shared and others are not shared. When the prediction of future returns for different stocks are viewed as different tasks, sharing some parameters across stocks is a form of multi-task learning. In a series of experiments with Canadian stocks, we obtain yearly returns that are more than 14\% above various benchmarks.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2414 topics={MultiTask,Finance},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2415 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2416
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2417 @TECHREPORT{Gingras-asynchronous-TR96,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2418 author = {Gingras, Fran{\c c}ois and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2419 title = {Handling asynchronous or missing financial data with recurrent networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2420 number = {1020},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2421 year = {1996},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2422 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2423 topics={Finance,Missing},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2424 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2425
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2426 @TECHREPORT{Gingras-financial-TR99,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2427 author = {Gingras, Fran{\c c}ois and Bengio, Yoshua and Nadeau, Claude},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2428 title = {On Out-of-Sample Statistics for Financial Time-Series},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2429 number = {2585},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2430 year = {1999},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2431 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2432 topics={Comparative,Finance},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2433 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2434
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2435 @INPROCEEDINGS{gingras2000,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2436 author = {Gingras, Fran{\c c}ois and Bengio, Yoshua and Nadeau, Claude},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2437 title = {On Out-of-Sample Statistics for Time-Series},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2438 booktitle = {Computational Finance 2000},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2439 year = {2000},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2440 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/out-err-cf2000.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2441 abstract = {This paper studies an out-of-sample statistic for time-series prediction that is analogous to the widely used R2 in-sample statistic. We propose and study methods to estimate the variance of this out-of-sample statistic. We suggest that the out-of-sample statistic is more robust to distributional and asymptotic assumptions behind many tests for in-sample statistics. Furthermore we argue that it may be more important in some cases to choose a model that generalizes as well as possible rather than choose the parameters that are closest to the true parameters. Comparative experiments are performed on a financial time-series (daily and monthly returns of the TSE300 index). The experiments are performed or varying prediction horizons and we study the relation between predictibility (out-of-sample R2), variability of the out-of-sample R2 statistic, and the prediction horizon.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2442 topics={Comparative,Finance},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2443 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2444
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2445 @INPROCEEDINGS{GlorotAISTATS2010,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2446 author = {Bengio, Yoshua and Glorot, Xavier},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2447 title = {Understanding the difficulty of training deep feedforward neural networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2448 booktitle = {Proceedings of AISTATS 2010},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2449 volume = {9},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2450 year = {2010},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2451 pages = {249-256},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2452 abstract = {Whereas before 2006 it appears that deep multi-layer neural networks were not successfully trained, since then several algorithms have been shown to successfully train them, with experimental results showing the superiority of deeper vs less deep architectures. All these experimental results were obtained with new initialization or training mechanisms. Our objective here is to understand better why standard gradient descent from random initialization is doing so poorly with deep neural networks, to better understand these recent relative successes and help design better algorithms in the future. We first observe the influence of the non-linear activations functions. We find that the logistic sigmoid activation is unsuited for deep networks with random initialization because of its mean value, which can drive especially the top hidden layer into saturation. Surprisingly, we find that saturated units can move out of saturation by themselves, albeit slowly, and explaining the plateaus sometimes seen when training neural networks. We find that a new non-linearity that saturates less can often be beneficial. Finally, we study how activations and gradients vary across layers and during training, with the idea that training may be more difficult when the singular values of the Jacobian associated with each layer are far from 1. Based on these considerations, we propose a new initialization scheme that brings substantially faster convergence.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2453 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2454
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2455 @INPROCEEDINGS{Gori89,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2456 author = {Gori, Marco and Bengio, Yoshua and De Mori, Renato},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2457 title = {BPS: a learning algorithm for capturing the dynamic nature of speech},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2458 booktitle = {International Joint Conference on Neural Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2459 volume = {2},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2460 year = {1989},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2461 pages = {417--424},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2462 publisher = {IEEE, New York},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2463 topics={Speech},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2464 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2465
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2466 @INCOLLECTION{Grandvalet+Bengio-ssl-2006,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2467 author = {Grandvalet, Yves and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2468 editor = {Chapelle, Olivier and {Sch{\"{o}}lkopf}, Bernhard and Zien, Alexander},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2469 title = {Entropy Regularization},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2470 booktitle = {Semi-Supervised Learning},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2471 year = {2006},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2472 pages = {151--168},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2473 publisher = {{MIT} Press},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2474 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/entropy_regularization_2006.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2475 abstract = {The problem of semi-supervised induction consists in learning a decision rule from
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2476 labeled and unlabeled data. This task can be undertaken by discriminative methods,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2477 provided that learning criteria are adapted consequently. In this chapter, we motivate the use of entropy regularization as a means to benefit from unlabeled data in
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2478 the framework of maximum a posteriori estimation. The learning criterion is derived
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2479 from clearly stated assumptions and can be applied to any smoothly parametrized
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2480 model of posterior probabilities. The regularization scheme favors low density separation, without any modeling of the density of input features. The contribution
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2481 of unlabeled data to the learning criterion induces local optima, but this problem
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2482 can be alleviated by deterministic annealing. For well-behaved models of posterior
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2483 probabilities, deterministic annealing {EM} provides a decomposition of the learning
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2484 problem in a series of concave subproblems. Other approaches to the semi-supervised
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2485 problem are shown to be close relatives or limiting cases of entropy regularization.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2486 A series of experiments illustrates the good behavior of the algorithm in terms of
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2487 performance and robustness with respect to the violation of the postulated low density separation assumption. The minimum entropy solution benefits from unlabeled
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2488 data and is able to challenge mixture models and manifold learning in a number of
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2489 situations.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2490 cat={B},topics={Unsupervised},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2491 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2492
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2493 @INPROCEEDINGS{graves+eck+schmidhuber:bio-adit2004,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2494 author = {Graves, A. and Eck, Douglas and Beringer, N. and Schmidhuber, Juergen},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2495 title = {Biologically Plausible Speech Recognition with {LSTM} Neural Nets},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2496 booktitle = {Proceedings of the First Int'l Workshop on Biologically Inspired Approaches to Advanced Information Technology (Bio-ADIT)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2497 year = {2004},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2498 pages = {127-136},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2499 url = {http://www.iro.umontreal.ca/~eckdoug/papers/2004_bioadit.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2500 abstract = {Long Short-Term Memory ({LSTM}) recurrent neural networks ({RNN}s) are local in space and time and closely related to a biological model of memory in the prefrontal cortex. Not only are they more biologically plausible than previous artificial {RNN}s, they also outperformed them on many artificially generated sequential processing tasks. This encouraged us to apply {LSTM} to more realistic problems, such as the recognition of spoken digits. Without any modification of the underlying algorithm, we achieved results comparable to state-of-the-art Hidden {Markov} Model ({HMM}) based recognisers on both the {TIDIGITS} and TI46 speech corpora. We conclude that {LSTM} should be further investigated as a biologically plausible basis for a bottom-up, neural net-based approach to speech recognition.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2501 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2502 sourcetype={Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2503 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2504
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2505 @TECHREPORT{graves+eck+schmidhuber:tr-digits2003,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2506 author = {Graves, A. and Eck, Douglas and Schmidhuber, Juergen},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2507 title = {Comparing {LSTM} Recurrent Networks and Spiking Recurrent Networks on the Recognition of Spoken Digits},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2508 number = {IDSIA-13-03},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2509 year = {2003},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2510 institution = {IDSIA},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2511 abstract = {One advantage of spiking recurrent neural networks ({SNN}s) is an ability to categorise data using a synchrony-based latching mechnanism. This is particularly useful in problems where timewarping is encountered, such as speech recognition. Differentiable recurrent neural networks ({RNN}s) by contrast fail at tasks involving difficult timewarping, despite having sequence learning capabilities superior to {SNN}s. In this paper we demonstrate that Long Short-Term Memory ({LSTM}) is an {RNN} capable of robustly categorizing timewarped speech data, thus combining the most useful features of both paradigms. We compare its performance to {SNN}s on two variants of a spoken digit identification task, using data from an international competition. The first task (described in Nature (Nadis 2003)) required the categorisation of spoken digits with only a single training exemplar, and was specifically designed to test robustness to timewarping. Here {LSTM} performed better than all the {SNN}s in the competition. The second task was to predict spoken digits using a larger training set. Here {LSTM} greatly outperformed an {SNN}-like model found in the literature. These results suggest that {LSTM} has a place in domains that require the learning of large timewarped datasets, such as automatic speech recognition.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2512 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2513 sourcetype={TechReport},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2514 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2515
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2516 @INPROCEEDINGS{haffner-98,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2517 author = {Haffner, Patrick and Bottou, {L{\'{e}}on} and G. Howard, Paul and Simard, Patrice and Bengio, Yoshua and {LeCun}, Yann},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2518 title = {Browsing through High Quality Document Images with {DjVu}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2519 booktitle = {Proc. of Advances in Digital Libraries 98},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2520 year = {1998},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2521 pages = {309--318},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2522 publisher = {IEEE},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2523 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/haffner-98.ps.gz},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2524 topics={HighDimensional},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2525 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2526
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2527 @INPROCEEDINGS{Hamel+al-2009,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2528 author = {Hamel, Philippe and Wood, Sean and Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2529 title = {Automatic Identification of Instrument Classes in Polyphonic and Poly-Instrument Audio},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2530 booktitle = {10th International Society for Music Information Retrieval Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2531 year = {2009},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2532 pages = {399--404},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2533 url = {http://ismir2009.ismir.net/proceedings/PS3-2.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2534 abstract = {We present and compare several models for automatic identiﬁcation of instrument classes in polyphonic and poly-instrument audio. The goal is to be able to identify which categories of instrument (Strings, Woodwind, Guitar, Piano, etc.) are present in a given audio example. We use a machine learning approach to solve this task. We constructed a system to generate a large database of musically relevant poly-instrument audio. Our database is generated from hundreds of instruments classiﬁed in 7 categories. Musical audio examples are generated by mixing multi-track MIDI ﬁles with thousands of instrument combinations. We compare three different classiﬁers : a Support Vector Machine ({SVM}), a Multilayer Perceptron (MLP) and a Deep Belief Network (DBN). We show that the DBN tends to outperform both the {SVM} and the MLP in most cases.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2535 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2536
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2537 @MISC{Hugo+al-snowbird-2007,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2538 author = {Larochelle, Hugo and Bengio, Yoshua and Erhan, Dumitru},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2539 title = {Generalization to a zero-data task: an empirical study},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2540 year = {2007},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2541 howpublished = {Talk and poster presented at the Learning Workshop(Snowbird), San Juan, Puerto Rico, 2007}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2542 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2543
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2544 @INPROCEEDINGS{hyper:2000:ijcnn,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2545 author = {Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2546 title = {Continuous Optimization of Hyper-Parameters},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2547 booktitle = {International Joint Conference on Neural Networks 2000},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2548 volume = {I},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2549 year = {2000},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2550 pages = {305--310},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2551 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/hyper-ijcnn2000.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2552 abstract = {Many machine learning algorithms can be formulated as the minimization of a training criterion which involves a hyper-parameter. This hyper-parameter is usually chosen by trial and error with a model selection criterion. In this paper we present a methodology to optimize several hyper-parameters, based on the computation of the gradient of a model selection criterion with respect to the hyper-parameters. In the case of a quadratic training criterion, the gradient of the selection criterion with respect to the hyper-parameters is efficiently computed by back-propagating through a Cholesky decomposition. In the more general case, we show that the implicit function theorem can be used to derive a formula for the hyper-parameter gradient involving second derivatives of the training criterion.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2553 topics={ModelSelection},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2554 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2555
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2556 @INPROCEEDINGS{ICML01,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2557 editor = {Brodley, Carla E. and Danyluk, Andrea Pohoreckyj},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2558 title = {Proceedings of the Eighteenth International Conference on Machine Learning (ICML'01)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2559 booktitle = {Proceedings of the Eighteenth International Conference on Machine Learning (ICML'01)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2560 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2561 publisher = {Morgan Kaufmann}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2562 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2563
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2564 @INPROCEEDINGS{ICML01-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2565 editor = {Brodley, Carla E. and Danyluk, Andrea Pohoreckyj},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2566 title = {Proceedings of the Eighteenth International Conference on Machine Learning (ICML'01)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2567 booktitle = {ICML'01},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2568 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2569 publisher = {Morgan Kaufmann}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2570 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2571
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2572
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2573 @INPROCEEDINGS{ICML02,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2574 editor = {Sammut, Claude and Hoffmann, Achim G.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2575 title = {Proceedings of the Nineteenth International Conference on Machine Learning (ICML'02)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2576 booktitle = {Proceedings of the Nineteenth International Conference on Machine Learning (ICML'02)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2577 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2578 publisher = {Morgan Kaufmann}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2579 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2580
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2581 @INPROCEEDINGS{ICML02-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2582 editor = {Sammut, Claude and Hoffmann, Achim G.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2583 title = {Proceedings of the Nineteenth International Conference on Machine Learning (ICML'02)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2584 booktitle = {ICML'02},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2585 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2586 publisher = {Morgan Kaufmann}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2587 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2588
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2589
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2590 @INPROCEEDINGS{ICML03,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2591 editor = {Fawcett, Tom and Mishra, Nina},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2592 title = {Proceedings of the Twenty International Conference on Machine Learning (ICML'03)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2593 booktitle = {Proceedings of the Twenty International Conference on Machine Learning (ICML'03)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2594 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2595 publisher = {AAAI Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2596 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2597
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2598 @INPROCEEDINGS{ICML03-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2599 editor = {Fawcett, Tom and Mishra, Nina},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2600 title = {Proceedings of the Twenty International Conference on Machine Learning (ICML'03)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2601 booktitle = {ICML'03},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2602 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2603 publisher = {AAAI Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2604 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2605
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2606
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2607 @INPROCEEDINGS{ICML04,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2608 editor = {Brodley, Carla E.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2609 title = {Proceedings of the Twenty-first International Conference on Machine Learning (ICML'04)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2610 booktitle = {Proceedings of the Twenty-first International Conference on Machine Learning (ICML'04)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2611 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2612 publisher = {ACM}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2613 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2614
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2615 @INPROCEEDINGS{ICML04-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2616 editor = {Brodley, Carla E.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2617 title = {Proceedings of the Twenty-first International Conference on Machine Learning (ICML'04)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2618 booktitle = {ICML'04},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2619 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2620 publisher = {ACM}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2621 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2622
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2623
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2624 @INPROCEEDINGS{ICML05-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2625 editor = {Raedt, Luc De and Wrobel, Stefan},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2626 title = {Proceedings of the Twenty-second International Conference on Machine Learning (ICML'05)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2627 booktitle = {ICML'05},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2628 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2629 publisher = {ACM}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2630 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2631
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2632
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2633 @INPROCEEDINGS{ICML06-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2634 editor = {Cohen, William W. and Moore, Andrew},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2635 title = {Proceedings of the Twenty-three International Conference on Machine Learning (ICML'06)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2636 booktitle = {ICML'06},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2637 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2638 publisher = {ACM}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2639 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2640
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2641
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2642 @INPROCEEDINGS{ICML07-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2643 editor = {Ghahramani, Zoubin},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2644 title = {Proceedings of the 24th International Conference on Machine Learning (ICML'07)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2645 booktitle = {ICML'07},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2646 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2647 publisher = {ACM}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2648 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2649
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2650
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2651 @INPROCEEDINGS{ICML08-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2652 editor = {Cohen, William W. and McCallum, Andrew and Roweis, Sam T.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2653 title = {Proceedings of the Twenty-fifth International Conference on Machine Learning (ICML'08)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2654 booktitle = {ICML'08},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2655 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2656 publisher = {ACM}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2657 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2658
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2659
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2660 @INPROCEEDINGS{ICML09-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2661 editor = {Bottou, {L{\'{e}}on} and Littman, Michael},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2662 title = {Proceedings of the Twenty-sixth International Conference on Machine Learning (ICML'09)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2663 booktitle = {ICML'09},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2664 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2665 publisher = {ACM}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2666 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2667
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2668
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2669 @INPROCEEDINGS{ICML96,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2670 editor = {Saitta, L.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2671 title = {Proceedings of the Thirteenth International Conference on Machine Learning (ICML'96)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2672 booktitle = {Proceedings of the Thirteenth International Conference on Machine Learning (ICML'96)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2673 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2674 publisher = {Morgan Kaufmann}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2675 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2676
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2677 @INPROCEEDINGS{ICML96-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2678 editor = {Saitta, L.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2679 title = {Proceedings of the Thirteenth International Conference on Machine Learning (ICML'96)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2680 booktitle = {ICML'96},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2681 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2682 publisher = {Morgan Kaufmann}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2683 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2684
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2685
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2686 @INPROCEEDINGS{ICML97,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2687 editor = {Fisher, Douglas H.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2688 title = {{}Proceedings of the Fourteenth International Conference on Machine Learning (ICML'97)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2689 booktitle = {Proceedings of the Fourteenth International Conference on Machine Learning (ICML'97)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2690 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2691 publisher = {Morgan Kaufmann}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2692 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2693
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2694 @INPROCEEDINGS{ICML97-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2695 editor = {Fisher, Douglas H.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2696 title = {{}Proceedings of the Fourteenth International Conference on Machine Learning (ICML'97)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2697 booktitle = {ICML'97},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2698 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2699 publisher = {Morgan Kaufmann}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2700 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2701
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2702
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2703 @INPROCEEDINGS{ICML98,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2704 editor = {Shavlik, Jude W.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2705 title = {Proceedings of the Fifteenth International Conference on Machine Learning (ICML'98)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2706 booktitle = {Proceedings of the Fifteenth International Conference on Machine Learning (ICML'98)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2707 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2708 publisher = {Morgan Kaufmann}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2709 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2710
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2711 @INPROCEEDINGS{ICML98-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2712 editor = {Shavlik, Jude W.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2713 title = {Proceedings of the Fifteenth International Conference on Machine Learning (ICML'98)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2714 booktitle = {ICML'98},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2715 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2716 publisher = {Morgan Kaufmann}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2717 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2718
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2719
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2720 @INPROCEEDINGS{ICML99,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2721 editor = {Bratko, Ivan and Dzeroski, Saso},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2722 title = {Proceedings of the Sixteenth International Conference on Machine Learning (ICML'99)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2723 booktitle = {Proceedings of the Sixteenth International Conference on Machine Learning (ICML'99)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2724 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2725 publisher = {Morgan Kaufmann}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2726 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2727
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2728 @INPROCEEDINGS{ICML99-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2729 editor = {Bratko, Ivan and Dzeroski, Saso},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2730 title = {Proceedings of the Sixteenth International Conference on Machine Learning (ICML'99)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2731 booktitle = {ICML'99},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2732 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2733 publisher = {Morgan Kaufmann}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2734 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2735
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2736
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2737 @INCOLLECTION{jaeger+eck:2007,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2738 author = {Jaeger, H. and Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2739 title = {Can't get you out of my head: {A} connectionist model of cyclic rehearsal},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2740 booktitle = {Modeling Communications with Robots and Virtual Humans},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2741 series = {{LNCS}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2742 year = {2007},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2743 publisher = {Springer-Verlag},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2744 url = {http://www.iro.umontreal.ca/~eckdoug/papers/2007_jaeger_eck.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2745 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2746 sourcetype={Chapter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2747 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2748
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2749 @MISC{James+al-snowbird-2008,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2750 author = {Bergstra, James and Bengio, Yoshua and Louradour, Jerome},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2751 title = {Image Classification using Higher-Order Neural Models},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2752 year = {2008},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2753 howpublished = {The Learning Workshop (Snowbird, Utah)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2754 url = {http://snowbird.djvuzone.org/2007/abstracts/161.pdf}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2755 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2756
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2757
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2758 @INPROCEEDINGS{Kegl+Bertin+Eck-2008,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2759 author = {K{\'{e}}gl, Bal{\'{a}}zs and Bertin-Mahieux, Thierry and Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2760 title = {Metropolis-Hastings Sampling in a FilterBoost Music Classifier},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2761 booktitle = {Music and machine learning workshop (ICML08)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2762 year = {2008}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2763 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2764
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2765 @INPROCEEDINGS{kegl2005b,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2766 author = {K{\'{e}}gl, Bal{\'{a}}zs},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2767 title = {Generalization Error and Algorithmic Convergence of Median Boosting.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2768 year = {2005},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2769 crossref = {NIPS17-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2770 abstract = {We have recently proposed an extension of ADABOOST to regression that uses the median of the base regressors as the final regressor. In this paper we extend theoretical results obtained for ADABOOST to median boosting and to its localized variant. First, we extend recent results on efficient margin maximizing to show that the algorithm can converge to the maximum achievable margin within a preset precision in a finite number of steps. Then we provide confidence-interval-type bounds on the generalization error.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2771 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2772
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2773 @ARTICLE{lacoste+eck:eurasip,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2774 author = {Lacoste, Alexandre and Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2775 title = {A Supervised Classification Algorithm For Note Onset Detection},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2776 journal = {EURASIP Journal on Applied Signal Processing},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2777 volume = {2007},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2778 number = {ID 43745},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2779 year = {2007},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2780 pages = {1--13},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2781 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2782 sourcetype={Journal},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2783 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2784
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2785 @MASTERSTHESIS{Lajoie2009,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2786 author = {Lajoie, Isabelle},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2787 keywords = {apprentissage non-supervis{\'{e}}, architecture profonde, auto-encodeur d{\'{e}}bruiteur, machine de {Boltzmann} restreinte, r{\'{e}}seau de neurones artiﬁciel},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2788 title = {Apprentissage de repr{\'{e}}sentations sur-compl{\`{e}}tes par entra{\^{\i}}nement d’auto-encodeurs},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2789 year = {2009},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2790 school = {Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2791 abstract = {Les avanc{\'{e}}s dans le domaine de l’intelligence artiﬁcielle, permettent {\`{a}} des syst{\`{e}}mes
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2792 informatiques de r{\'{e}}soudre des t{\^{a}}ches de plus en plus complexes li{\'{e}}es par exemple {\`{a}}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2793 la vision, {\`{a}} la compr{\'{e}}hension de signaux sonores ou au traitement de la langue. Parmi
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2794 les mod{\`{e}}les existants, on retrouve les R{\'{e}}seaux de Neurones Artiﬁciels (RNA), dont la
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2795 popularit{\'{e}} a fait un grand bond en avant avec la d{\'{e}}couverte de Hinton et al. [22], soit
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2796 l’utilisation de Machines de {Boltzmann} Restreintes (RBM) pour un pr{\'{e}}-entra{\^{\i}}nement
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2797 non-supervis{\'{e}} couche apr{\`{e}}s couche, facilitant grandement l’entra{\^{\i}}nement supervis{\'{e}} du
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2798 r{\'{e}}seau {\`{a}} plusieurs couches cach{\'{e}}es (DBN), entra{\^{\i}}nement qui s’av{\'{e}}rait jusqu’alors tr{\`{e}}s
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2799 difﬁcile {\`{a}} r{\'{e}}ussir. Depuis cette d{\'{e}}couverte, des chercheurs ont {\'{e}}tudi{\'{e}} l’efﬁcacit{\'{e}} de nouvelles strat{\'{e}}gies de pr{\'{e}}-entra{\^{\i}}nement, telles que l’empilement d’auto-encodeurs traditionnels (SAE) [5, 38], et l’empilement d’auto-encodeur d{\'{e}}bruiteur (SDAE) [44].
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2800 C’est dans ce contexte qu’a d{\'{e}}but{\'{e}} la pr{\'{e}}sente {\'{e}}tude. Apr{\`{e}}s un bref passage en revue des notions de base du domaine de l’apprentissage machine et des m{\'{e}}thodes de
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2801 pr{\'{e}}-entra{\^{\i}}nement employ{\'{e}}es jusqu’{\`{a}} pr{\'{e}}sent avec les modules RBM, AE et DAE, nous
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2802 avons approfondi notre compr{\'{e}}hension du pr{\'{e}}-entra{\^{\i}}nement de type SDAE, explor{\'{e}} ses
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2803 diff{\'{e}}rentes propri{\'{e}}t{\'{e}}s et {\'{e}}tudi{\'{e}} des variantes de SDAE comme strat{\'{e}}gie d’initialisation
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2804 d’architecture profonde. Nous avons ainsi pu, entre autres choses, mettre en lumi{\`{e}}re
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2805 l’inﬂuence du niveau de bruit, du nombre de couches et du nombre d’unit{\'{e}}s cach{\'{e}}es
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2806 sur l’erreur de g{\'{e}}n{\'{e}}ralisation du SDAE. Nous avons constat{\'{e}} une am{\'{e}}lioration de la
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2807 performance sur la t{\^{a}}che supervis{\'{e}}e avec l’utilisation des bruits poivre et sel (PS) et
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2808 gaussien (GS), bruits s’av{\'{e}}rant mieux justiﬁ{\'{e}}s que celui utilis{\'{e}} jusqu’{\`{a}} pr{\'{e}}sent, soit le
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2809 masque {\`{a}} z{\'{e}}ro (MN). De plus, nous avons d{\'{e}}montr{\'{e}} que la performance proﬁtait d’une
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2810 emphase impos{\'{e}}e sur la reconstruction des donn{\'{e}}es corrompues durant l’entra{\^{\i}}nement
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2811 des diff{\'{e}}rents DAE. Nos travaux ont aussi permis de r{\'{e}}v{\'{e}}ler que le DAE {\'{e}}tait en mesure d’apprendre, sur des images naturelles, des ﬁltres semblables {\`{a}} ceux retrouv{\'{e}}s dans
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2812 les cellules V1 du cortex visuel, soit des ﬁltres d{\'{e}}tecteurs de bordures. Nous aurons par
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2813 ailleurs pu montrer que les repr{\'{e}}sentations apprises du SDAE, compos{\'{e}}es des caract{\'{e}}ristiques ainsi extraites, s’av{\'{e}}raient fort utiles {\`{a}} l’apprentissage d’une machine {\`{a}} vecteurs de
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2814 support ({SVM}) lin{\'{e}}aire ou {\`{a}} noyau gaussien, am{\'{e}}liorant grandement sa performance de
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2815 g{\'{e}}n{\'{e}}ralisation. Aussi, nous aurons observ{\'{e}} que similairement au DBN, et contrairement
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2816 au SAE, le SDAE poss{\'{e}}dait une bonne capacit{\'{e}} en tant que mod{\`{e}}le g{\'{e}}n{\'{e}}rateur. Nous
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2817 avons {\'{e}}galement ouvert la porte {\`{a}} de nouvelles strat{\'{e}}gies de pr{\'{e}}-entra{\^{\i}}nement et d{\'{e}}couvert le potentiel de l’une d’entre elles, soit l’empilement d’auto-encodeurs rebruiteurs
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2818 (SRAE).}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2819 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2820
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2821 @INPROCEEDINGS{lamere+eck:ismir2007,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2822 author = {Lamere, Paul and Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2823 editor = {},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2824 title = {Using 3D Visualizations to Explore and Discover Music},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2825 booktitle = {{Proceedings of the 8th International Conference on Music Information Retrieval ({ISMIR} 2007)}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2826 year = {2007},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2827 publisher = {},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2828 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2829 sourcetype={Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2830 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2831
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2832 @ARTICLE{Larochelle+al-2010,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2833 author = {Larochelle, Hugo and Bengio, Yoshua and Turian, Joseph},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2834 title = {Tractable Multivariate Binary Density Estimation and the Restricted {Boltzmann} Forest},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2835 journal = {Neural Computation},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2836 year = {2010},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2837 note = {To appear}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2838 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2839
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2840 @INPROCEEDINGS{Larochelle+Bengio-2008,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2841 author = {Larochelle, Hugo and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2842 title = {Classification using Discriminative Restricted {B}oltzmann Machines},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2843 year = {2008},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2844 pages = {536--543},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2845 crossref = {ICML08-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2846 abstract = {Recently, many applications for Restricted {Boltzmann} Machines (RBMs) have been developed for a large variety of learning problems. However, RBMs are usually used as feature extractors for another learning algorithm or to provide a good initialization
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2847 for deep feed-forward neural network classifiers, and are not considered as a standalone solution to classification problems. In
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2848 this paper, we argue that RBMs provide a self-contained framework for deriving competitive non-linear classifiers. We present an evaluation of different learning algorithms for
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2849 RBMs which aim at introducing a discriminative component to RBM training and improve their performance as classifiers. This
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2850 approach is simple in that RBMs are used directly to build a classifier, rather than as a stepping stone. Finally, we demonstrate how discriminative RBMs can also be successfully employed in a semi-supervised setting.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2851 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2852
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2853 @INPROCEEDINGS{Larochelle-2009,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2854 author = {Larochelle, Hugo and Erhan, Dumitru and Vincent, Pascal},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2855 title = {Deep Learning using Robust Interdependent Codes},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2856 booktitle = {Proceedings of the Twelfth International Conference on Artificial Intelligence and Statistics (AISTATS 2009)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2857 year = {2009},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2858 pages = {312--319},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2859 date = "April 16-18, 2009",
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2860 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2861
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2862 @ARTICLE{Larochelle-jmlr-2009,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2863 author = {Larochelle, Hugo and Bengio, Yoshua and Louradour, Jerome and Lamblin, Pascal},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2864 title = {Exploring Strategies for Training Deep Neural Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2865 volume = {10},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2866 year = {2009},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2867 pages = {1--40},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2868 journal = {Journal of Machine Learning Research},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2869 abstract = {Deep multi-layer neural networks have many levels of non-linearities allowing them to compactly represent highly non-linear and highly-varying functions. However, until recently it was not clear how to train such deep networks, since gradient-based optimization starting from random initialization often appears to get stuck in poor solutions. Hinton et al. recently proposed a greedy layer-wise unsupervised learning procedure relying on the training algorithm of restricted {Boltzmann} machines (RBM) to initialize the parameters of a deep belief network (DBN), a generative model with many layers of hidden causal variables. This was followed by the proposal of another greedy layer-wise procedure, relying on the usage of autoassociator networks. In the context of the above optimization problem, we study these algorithms empirically to better understand their success. Our experiments confirm the hypothesis that the greedy layer-wise unsupervised training strategy helps the optimization by initializing weights in a region near a good local minimum, but also implicitly acts as a sort of regularization that brings better generalization and encourages internal distributed representations that are high-level abstractions of the input. We also present a series of experiments aimed at evaluating the link between the performance of deep neural networks and practical aspects of their topology, for example, demonstrating cases where the addition of more depth helps. Finally, we empirically explore simple variants of these training algorithms, such as the use of different RBM input unit distributions, a simple way of combining gradient estimators to improve performance, as well as on-line versions of those algorithms.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2870 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2871
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2872 @PHDTHESIS{Larochelle-PhD-2009,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2873 author = {Larochelle, Hugo},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2874 keywords = {apprentissage non-supervis{\'{e}}, architecture profonde, autoassociateur, autoencodeur, machine de {Boltzmann} restreinte, r{\'{e}}seau de neurones artificiel},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2875 title = {{\'{E}}tude de techniques d'apprentissage non-supervis{\'{e}} pour l'am{\'{e}}lioration de l'entra{\^{\i}}nement supervis{\'{e}} de mod{\`{e}}les connexionnistes},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2876 year = {2009},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2877 school = {University of Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2878 abstract = {Le domaine de l'intelligence artificielle a pour objectif le d{\'{e}}veloppement de syst{\`{e}}mes informatiques capables de simuler des comportements normalement associ{\'{e}}s {\`{a}} l'intelligence humaine. On aimerait entre autres pouvoir construire une machine qui puisse
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2879 r{\'{e}}soudre des t{\^{a}}ches li{\'{e}}es {\`{a}} la vision (e.g., la reconnaissance d'objet), au traitement de la langue (e.g., l'identification du sujet d'un texte) ou au traitement de signaux sonores (e.g., la reconnaissance de la parole).
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2880 Une approche d{\'{e}}velopp{\'{e}}e afin de r{\'{e}}soudre ce genre de t{\^{a}}ches est bas{\'{e}}e sur l'apprentissage automatique de mod{\`{e}}les {\`{a}} partir de donn{\'{e}}es {\'{e}}tiquet{\'{e}}es refl{\'{e}}tant le comportement intelligent {\`{a}} {\'{e}}muler. Entre autre, il a {\'{e}}t{\'{e}} propos{\'{e}} de mod{\'{e}}liser le calcul n{\'{e}}cessaire {\`{a}} la
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2881 r{\'{e}}solution d'une t{\^{a}}che {\`{a}} l'aide d'un r{\'{e}}seau de neurones artificiel, dont il est possible d'adapter le comportement {\`{a}} l'aide de la r{\'{e}}tropropagation [99, 131] d'un gradient informatif sur les erreurs commises par le r{\'{e}}seau. Populaire durant les ann{\'{e}}es 80, cette
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2882 approche sp{\'{e}}cifique a depuis perdu partiellement de son attrait, suite au d{\'{e}}veloppement des m{\'{e}}thodes {\`{a}} noyau. Celles-ci sont souvent plus stables, plus faciles {\`{a}} utiliser et leur performance est souvent au moins aussi {\'{e}}lev{\'{e}}e pour une vaste gamme de probl{\`{e}}mes.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2883 Les m{\'{e}}thodes d'apprentissage automatique ont donc progress{\'{e}} dans leur fonctionnement, mais aussi dans la complexit{\'{e}} des probl{\`{e}}mes auxquels elles se sont attaqu{\'{e}}. Ainsi, plus r{\'{e}}cemment, des travaux [12, 15] ont commenc{\'{e}} {\`{a}} {\'{e}}mettre des doutes sur la capacit{\'{e}} des machines {\`{a}} noyau {\`{a}} pouvoir efficacement r{\'{e}}soudre des probl{\`{e}}mes de la complexit{\'{e}} requise par l'intelligence artificielle. Parall{\`{e}}lement, Hinton et al. [81] faisaient une perc{\'{e}}e dans l'apprentissage automatique de r{\'{e}}seaux de neurones, en proposant une proc{\'{e}}dure permettant l'entra{\^{\i}}nement de r{\'{e}}seaux de neurones d'une plus grande complexit{\'{e}} (i.e., avec plus de couches de neurones cach{\'{e}}es) qu'il n'{\'{e}}tait possible auparavant.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2884 C'est dans ce contexte qu'ont {\'{e}}t{\'{e}} conduits les travaux de cette th{\`{e}}se. Cette th{\`{e}}se d{\'{e}}bute par une exposition des principes de base de l'apprentissage automatique (chapitre 1) et une discussion des obstacles {\`{a}} l'obtention d'un mod{\`{e}}le ayant une bonne performance
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2885 de g{\'{e}}n{\'{e}}ralisation (chapitre 2). Puis, sont pr{\'{e}}sent{\'{e}}es les contributions apport{\'{e}}es dans le cadre de cinq articles, contributions qui sont toutes bas{\'{e}}es sur l'utilisation d'une certaine
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2886 forme d'apprentissage non-supervis{\'{e}}.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2887 Le premier article (chapitre 4) propose une m{\'{e}}thode d'entra{\^{\i}}nement pour un type sp{\'{e}}cifique de r{\'{e}}seau {\`{a}} une seule couche cach{\'{e}}e (la machine de {Boltzmann} restreinte) bas{\'{e}}e sur une combinaison des apprentissages supervis{\'{e}} et non-supervis{\'{e}}. Cette m{\'{e}}thode permet d'obtenir une meilleure performance de g{\'{e}}n{\'{e}}ralisation qu'un r{\'{e}}seau de neurones standard ou qu'une machine {\`{a}} vecteurs de support {\`{a}} noyau, et met en {\'{e}}vidence de fa{\c c}on
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2888 explicite les b{\'{e}}n{\'{e}}fices qu'apporte l'apprentissage non-supervis{\'{e}} {\`{a}} l'entra{\^{\i}}nement d'un r{\'{e}}seau de neurones.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2889 Ensuite, dans le second article (chapitre 6), on {\'{e}}tudie et {\'{e}}tend la proc{\'{e}}dure d'entra{\^{\i}}nement propos{\'{e}}e par Hinton et al. [81]. Plus sp{\'{e}}cifiquement, on y propose une approche diff{\'{e}}rente mais plus flexible pour initialiser un r{\'{e}}seau {\`{a}} plusieurs couches cach{\'{e}}es, bas{\'{e}}e sur un r{\'{e}}seau autoassociateur. On y explore aussi l'impact du nombre de couches et de neurones par couche sur la performance d'un r{\'{e}}seau et on y d{\'{e}}crit diff{\'{e}}rentes variantes mieux adapt{\'{e}}es {\`{a}} l'apprentissage en ligne ou pour donn{\'{e}}es {\`{a}} valeurs continues.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2890 Dans le troisi{\`{e}}me article (chapitre 8), on explore plut{\^{o}}t la performance de r{\'{e}}seaux profonds sur plusieurs probl{\`{e}}mes de classification diff{\'{e}}rents. Les probl{\`{e}}mes choisis ont la propri{\'{e}}t{\'{e}} d'avoir {\'{e}}t{\'{e}} g{\'{e}}n{\'{e}}r{\'{e}}s {\`{a}} partir de plusieurs facteurs de variation. Cette propri{\'{e}}t{\'{e}}, qui caract{\'{e}}rise les probl{\`{e}}mes li{\'{e}}s {\`{a}} l'intelligence artificielle, pose difficult{\'{e}} aux machines {\`{a}} noyau, tel que confirm{\'{e}} par les exp{\'{e}}riences de cet article.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2891 Le quatri{\`{e}}me article (chapitre 10) pr{\'{e}}sente une am{\'{e}}lioration de l'approche bas{\'{e}}e sur les r{\'{e}}seaux autoassociateurs. Cette am{\'{e}}lioration applique une modification simple {\`{a}} la proc{\'{e}}dure d'entra{\^{\i}}nement d'un r{\'{e}}seau autoassociateur, en « bruitant » les entr{\'{e}}es du r{\'{e}}seau afin que celui-ci soit forc{\'{e}} {\`{a}} la d{\'{e}}bruiter.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2892 Le cinqui{\`{e}}me et dernier article (chapitre 12) apporte une autre am{\'{e}}lioration aux r{\'{e}}seaux autoassociateurs, en permettant des interactions d'inhibition ou d'excitation entre les neurones cach{\'{e}}s de ces r{\'{e}}seaux. On y d{\'{e}}montre que de telles interactions peuvent
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2893 {\^{e}}tre apprises et sont b{\'{e}}n{\'{e}}fiques {\`{a}} la performance d'un r{\'{e}}seau profond.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2894 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2895
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2896 @INPROCEEDINGS{Larochelle2008,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2897 author = {Larochelle, Hugo and Erhan, Dumitru and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2898 title = {Zero-data Learning of New Tasks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2899 booktitle = {AAAI Conference on Artificial Intelligence},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2900 year = {2008},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2901 url = {http://www-etud.iro.umontreal.ca/~larocheh/publications/aaai2008_zero-data.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2902 abstract = {Recently, many applications for Restricted {Boltzmann} Machines (RBMs) have been developed for a large variety of learning problems. However, RBMs are usually used as feature extractors for another learning algorithm or to provide a good initialization
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2903 for deep feed-forward neural network classifiers, and are not considered as a standalone solution to classification problems. In
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2904 this paper, we argue that RBMs provide a self-contained framework for deriving competitive non-linear classifiers. We present an evaluation of different learning algorithms for
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2905 RBMs which aim at introducing a discriminative component to RBM training and improve their performance as classifiers. This
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2906 approach is simple in that RBMs are used directly to build a classifier, rather than as a stepping stone. Finally, we demonstrate how discriminative RBMs can also be successfully employed in a semi-supervised setting.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2907 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2908
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2909 @INPROCEEDINGS{LarochelleH2007,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2910 author = {Larochelle, Hugo and Erhan, Dumitru and Courville, Aaron and Bergstra, James and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2911 title = {An Empirical Evaluation of Deep Architectures on Problems with Many Factors of Variation},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2912 year = {2007},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2913 pages = {473--480},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2914 crossref = {ICML07-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2915 abstract = {Recently, several learning algorithms relying on models with deep architectures have been proposed. Though they have demonstrated impressive performance, to date, they have only been evaluated on relatively simple problems such as digit recognition in a controlled environment, for which many machine learning algorithms already report reasonable results. Here, we present a series of experiments which indicate that these models show promise in solving harder learning problems that exhibit many factors of variation. These models are compared with well-established algorithms such as Support Vector Machines and single hidden-layer feed-forward neural networks.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2916 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2917
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2918 @MASTERSTHESIS{Latendresse-MSc,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2919 author = {Latendresse, Simon},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2920 title = {L'utilisation d'hyper-param{\`{e}}tres pour la selection de variables},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2921 year = {1999},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2922 school = {Universit{\'{e}} de Montreal, Dept. IRO},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2923 note = {(in French)}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2924 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2925
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2926 @MASTERSTHESIS{Lauzon99,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2927 author = {Lauzon, Vincent-Philippe},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2928 title = {Mod{\'{e}}les statistiques comme algorithmes d'apprentissage et {MMCC}s; pr{\'{e}}diction de s{\'{e}}ries financi{\`{e}}res},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2929 year = {1999},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2930 school = {D{\'{e}}epartement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2931 crossref = {DIRO}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2932 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2933
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2934 @INPROCEEDINGS{lecun-93,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2935 author = {{LeCun}, Yann and Bengio, Yoshua and Henderson, Donnie and Weisbuch, A. and Weissman, H. and L., Jackel},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2936 title = {On-line handwriting recognition with neural networks: spatial representation versus temporal representation.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2937 booktitle = {Proc. International Conference on handwriting and drawing.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2938 year = {1993},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2939 publisher = {Ecole Nationale Superieure des Telecommunications},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2940 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/lecun-93.ps.gz},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2941 topics={PriorKnowledge,Speech},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2942 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2943
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2944 @INPROCEEDINGS{lecun-99,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2945 author = {{LeCun}, Yann and Haffner, Patrick and Bottou, {L{\'{e}}on} and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2946 editor = {Forsyth, D.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2947 title = {Object Recognition with Gradient-Based Learning},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2948 booktitle = {Shape, Contour and Grouping in Computer Vision},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2949 year = {1999},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2950 pages = {319-345},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2951 publisher = {Springer},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2952 url = {orig/lecun-99.ps.gz},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2953 topics={PriorKnowledge,Speech},cat={B},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2954 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2955
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2956 @TECHREPORT{lecun-99b,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2957 author = {{LeCun}, Yann and Haffner, Patrick and Bottou, {L{\'{e}}on} and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2958 title = {Gradient-Based Learning for Object Detection, Segmentation and Recognition},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2959 year = {1999},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2960 institution = {AT\&T Labs},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2961 url = {orig/lecun-99b.ps.gz},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2962 topics={Speech},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2963 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2964
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2965 @INPROCEEDINGS{lecun-bengio-94,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2966 author = {{LeCun}, Yann and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2967 title = {Word-level training of a handwritten word recognizer based on convolutional neural networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2968 booktitle = {Proc. of the International Conference on Pattern Recognition},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2969 volume = {II},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2970 year = {1994},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2971 pages = {88--92},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2972 publisher = {IEEE},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2973 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/icpr-word.ps},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2974 abstract = {We introduce a new approach for on-line recognition of handwritten words written in unconstrained mixed style. Words are represented by low resolution annotated images where each pixel contains information about trajectory direction and curvature. The recognizer is a convolution network which can be spatially replicated. From the network output, a hidden {Markov} model produces word scores. The entire system is globally trained to minimize word-level errors.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2975 topics={Speech},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2976 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2977
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2978 @INPROCEEDINGS{lecun-bengio-95a,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2979 author = {{LeCun}, Yann and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2980 editor = {Arbib, M. A.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2981 title = {Convolutional Networks for Images, Speech, and Time-Series},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2982 booktitle = {The Handbook of Brain Theory and Neural Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2983 year = {1995},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2984 pages = {255--257},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2985 publisher = {MIT Press},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2986 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/handbook-convo.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2987 topics={PriorKnowledge,Speech},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2988 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2989
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2990 @INCOLLECTION{lecun-bengio-95b,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2991 author = {{LeCun}, Yann and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2992 editor = {Arbib, M. A.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2993 title = {Pattern Recognition and Neural Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2994 booktitle = {The Handbook of Brain Theory and Neural Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2995 year = {1995},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2996 pages = {711--714},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2997 publisher = {MIT Press},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2998 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/handbook-patrec.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	2999 topics={PriorKnowledge,Speech},cat={B},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3000 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3001
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3002 @ARTICLE{LeCun98,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3003 author = {{LeCun}, Yann and Bottou, {L{\'{e}}on} and Bengio, Yoshua and Haffner, Patrick},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3004 title = {Gradient-Based Learning Applied to Document Recognition},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3005 journal = {Proceedings of the IEEE},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3006 volume = {86},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3007 number = {11},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3008 year = {1998},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3009 pages = {2278--2324},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3010 abstract = {Multilayer Neural Networks trained with the backpropagation algorithm constitute the best example of a successful Gradient-Based Learning technique. Given an appropriate network architecture, Gradient-Based Learning algorithms can be used to synthesize a complex decision surface that can classify high-dimensional patterns such as handwritten characters, with minimal preprocessing. This paper reviews various methods applied to handwritten character recognition and compares them on a standard handwritten digit recognition task. Convolutional Neural Networks, that are specifically designed to deal with the variability of 2D shapes, are shown to outperform all other techniques.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3011 Real-life document recognition systems are composed or multiple modules including field extraction, segmentation, recognition, and language modeling. A new learning paradigm, called Graph Transformer Networks (GTN), allows such multi-module systems to be trained globally using Gradient-Based methods so as to minimize an overall performance measure.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3012 Two systems for on-line handwriting recognition are described. Experiments demonstrate the advantage of global training, and the flexibility of Graph Transformer Networks.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3013 A Graph Transformer Network for reading bank check is also described. It uses Convolutional Neural Network character recognizers combined with global training techniques to provides record accuracy on business and personal checks. It is deployed commercially and reads several million checks per day.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3014 topics={PriorKnowledge,Speech},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3015 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3016
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3017 @INPROCEEDINGS{Lecun_icassp97,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3018 author = {{LeCun}, Yann and Bottou, {L{\'{e}}on} and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3019 title = {Reading Checks with graph transformer networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3020 booktitle = {International Conference on Acoustics, Speech and Signal Processing},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3021 volume = {1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3022 year = {1997},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3023 pages = {151--154},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3024 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/lecun-bottou-bengio-97.ps.gz},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3025 topics={Speech},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3026 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3027
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3028 @ARTICLE{LeRoux+Bengio-2010,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3029 author = {Le Roux, Nicolas and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3030 title = {Deep Belief Networks are Compact Universal Approximators},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3031 journal = {Neural Computation},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3032 year = {2010},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3033 note = {To appear}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3034 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3035
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3036 @TECHREPORT{LeRoux-Bengio-2007-TR,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3037 author = {Le Roux, Nicolas and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3038 title = {Representational Power of Restricted {B}oltzmann Machines and Deep Belief Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3039 number = {1294},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3040 year = {2007},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3041 institution = {D{\'{e}}partement d'Informatique et de Recherche Op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3042 abstract = {Deep Belief Networks (DBN) are generative neural network models with
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3043 many layers of hidden explanatory factors, recently introduced by Hinton et al.,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3044 along with a greedy layer-wise unsupervised learning algorithm. The building
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3045 block of a DBN is a probabilistic model called a Restricted {Boltzmann} Machine
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3046 (RBM), used to represent one layer of the model. Restricted {Boltzmann} Machines
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3047 are interesting because inference is easy in them, and because they have been
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3048 successfully used as building blocks for training deeper models.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3049 We first prove that adding hidden units yields strictly improved modeling
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3050 power, while a second theorem shows that RBMs are universal approximators of
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3051 discrete distributions. We then study the question of whether DBNs with more
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3052 layers are strictly more powerful in terms of representational power. This
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3053 suggests a new and less greedy criterion for training RBMs within DBNs.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3054 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3055
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3056 @ARTICLE{LeRoux-Bengio-2008,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3057 author = {Le Roux, Nicolas and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3058 title = {Representational Power of Restricted {B}oltzmann Machines and Deep Belief Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3059 journal = {Neural Computation},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3060 volume = {20},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3061 number = {6},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3062 year = {2008},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3063 pages = {1631--1649},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3064 abstract = {Deep Belief Networks (DBN) are generative neural network models with many layers of hidden explanatory factors, recently introduced by Hinton et al., along with a greedy layer-wise unsupervised learning algorithm. The building block of a DBN is a probabilistic model called a Restricted {Boltzmann} Machine (RBM), used to represent one layer of the model. Restricted {Boltzmann} Machines are interesting because inference is easy in them, and because they have been successfully used as building blocks for training deeper models. We first prove that adding hidden units yields strictly improved modelling power, while a second theorem shows that RBMs are universal approximators of discrete distributions. We then study the question of whether DBNs with more layers are strictly more powerful in terms of representational power. This suggests a new and less greedy criterion for training RBMs within DBNs.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3065 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3066
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3067 @INPROCEEDINGS{LeRoux-continuous,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3068 author = {Le Roux, Nicolas and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3069 title = {Continuous Neural Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3070 booktitle = {Proceedings of the Eleventh International Conference on Artificial Intelligence and Statistics (AISTATS'07)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3071 year = {2007},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3072 publisher = {Omnipress},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3073 abstract = {This article extends neural networks to the case of an uncountable number of hidden units, in several ways. In the first approach proposed, a finite parametrization is possible, allowing gradient-based learning. While having the same number of parameters as an ordinary neural network, its internal structure suggests that it can represent some smooth functions much more compactly. Under mild assumptions, we also find better error bounds than with ordinary neural networks. Furthermore, this parametrization may help reducing the problem of saturation of the neurons. In a second approach, the input-to-hidden weights arefully non-parametric, yielding a kernel machine for which we demonstrate a simple kernel formula. Interestingly, the resulting kernel machine can be made hyperparameter-free and still generalizes in spite of an absence of explicit regularization.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3074 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3075
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3076 @PHDTHESIS{LeRoux-PhD-2008,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3077 author = {Le Roux, Nicolas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3078 title = {Avanc{\'{e}}es th{\'{e}}oriques sur la repr{\'{e}}sentation et l'optimisation des r{\'{e}}seaux de neurones},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3079 year = {2008},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3080 school = {Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3081 abstract = {Les r{\'{e}}seaux de neurones artificiels ont {\'{e}}t{\'{e}} abondamment utilis{\'{e}}s dans la communaut{\'{e}} de l'apprentissage machine depuis les ann{\'{e}}es 80. Bien qu'ils aient {\'{e}}t{\'{e}} {\'{e}}tudi{\'{e}}s pour la premi{\`{e}}re fois il y a cinquante ans par Rosenblatt [68], ils ne furent r{\'{e}}ellement populaires qu'apr{\`{e}}s l'apparition de la r{\'{e}}tropropagation du gradient, en 1986 [71].
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3082 En 1989, il a {\'{e}}t{\'{e}} prouv{\'{e}} [44] qu'une classe sp{\'{e}}cifique de r{\'{e}}seaux de neurones (les r{\'{e}}seaux de neurones {\`{a}} une couche cach{\'{e}}e) {\'{e}}tait suffisamment puissante pour pouvoir approximer presque n'importe quelle fonction avec une pr{\'{e}}cision arbitraire : le th{\'{e}}or{\`{e}}me d'approximation universelle. Toutefois, bien que ce th{\'{e}}or{\`{e}}me e{\^{u}}t pour cons{\'{e}}quence un int{\'{e}}r{\^{e}}t accru pour les r{\'{e}}seaux de neurones, il semblerait qu'aucun effort n'ait {\'{e}}t{\'{e}} fait pour profiter de cette propri{\'{e}}t{\'{e}}.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3083 En outre, l'optimisation des r{\'{e}}seaux de neurones {\`{a}} une couche cach{\'{e}}e n'est pas convexe. Cela a d{\'{e}}tourn{\'{e}} une grande partie de la communaut{\'{e}} vers d'autres algorithmes, comme par exemple les machines {\`{a}} noyau (machines {\`{a}} vecteurs de support et r{\'{e}}gression
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3084 {\`{a}} noyau, entre autres).
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3085 La premi{\`{e}}re partie de cette th{\`{e}}se pr{\'{e}}sentera les concepts d'apprentissage machine g{\'{e}}n{\'{e}}raux n{\'{e}}cessaires {\`{a}} la compr{\'{e}}hension des algorithmes utilis{\'{e}}s. La deuxi{\`{e}}me partie se focalisera plus sp{\'{e}}cifiquement sur les m{\'{e}}thodes {\`{a}} noyau et les r{\'{e}}seaux de neurones. La troisi{\`{e}}me partie de ce travail visera ensuite {\`{a}} {\'{e}}tudier les limitations des machines {\`{a}} noyaux et {\`{a}} comprendre les raisons pour lesquelles elles sont inadapt{\'{e}}es {\`{a}} certains probl{\`{e}}mes que nous avons {\`{a}} traiter.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3086 La quatri{\`{e}}me partie pr{\'{e}}sente une technique permettant d'optimiser les r{\'{e}}seaux de neurones {\`{a}} une couche cach{\'{e}}e de mani{\`{e}}re convexe. Bien que cette technique s'av{\`{e}}re difficilement exploitable pour des probl{\`{e}}mes de grande taille, une version approch{\'{e}}e permet d'obtenir une bonne solution dans un temps raisonnable.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3087 La cinqui{\`{e}}me partie se concentre sur les r{\'{e}}seaux de neurones {\`{a}} une couche cach{\'{e}}e infinie. Cela leur permet th{\'{e}}oriquement d'exploiter la propri{\'{e}}t{\'{e}} d'approximation universelle et ainsi d'approcher facilement une plus grande classe de fonctions.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3088 Toutefois, si ces deux variations sur les r{\'{e}}seaux de neurones {\`{a}} une couche cach{\'{e}}e leur conf{\`{e}}rent des propri{\'{e}}t{\'{e}}s int{\'{e}}ressantes, ces derniers ne peuvent extraire plus que des concepts de bas niveau. Les m{\'{e}}thodes {\`{a}} noyau souffrant des m{\^{e}}mes limites, aucun de
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3089 ces deux types d'algorithmes ne peut appr{\'{e}}hender des probl{\`{e}}mes faisant appel {\`{a}} l'apprentissage de concepts de haut niveau.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3090 R{\'{e}}cemment sont apparus les Deep Belief Networks [39] qui sont des r{\'{e}}seaux de neurones {\`{a}} plusieurs couches cach{\'{e}}es entra{\^{\i}}n{\'{e}}s de mani{\`{e}}re efficace. Cette profondeur leur permet d'extraire des concepts de haut niveau et donc de r{\'{e}}aliser des t{\^{a}}ches hors
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3091 de port{\'{e}}e des algorithmes conventionnels. La sixi{\`{e}}me partie {\'{e}}tudie des propri{\'{e}}t{\'{e}}s de ces r{\'{e}}seaux profonds.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3092 Les probl{\`{e}}mes que l'on rencontre actuellement n{\'{e}}cessitent non seulement des algorithmes capables d'extraire des concepts de haut niveau, mais {\'{e}}galement des m{\'{e}}thodes d'optimisation capables de traiter l'immense quantit{\'{e}} de donn{\'{e}}es parfois disponibles, si possible en temps r{\'{e}}el. La septi{\`{e}}me partie est donc la pr{\'{e}}sentation d'une nouvelle technique permettant une optimisation plus rapide.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3093 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3094
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3095 @ARTICLE{lheureux-04,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3096 author = {{L'Heureux}, Pierre-Jean and Carreau, Julie and Bengio, Yoshua and Delalleau, Olivier and Yue, Shi Yi},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3097 title = {Locally Linear Embedding for dimensionality reduction in {QSAR}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3098 journal = {Journal of Computer-Aided Molecular Design},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3099 volume = {18},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3100 year = {2004},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3101 pages = {475--482},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3102 abstract = {Current practice in Quantitative Structure Activity Relationship (QSAR) methods usually involves generating a great number of chemical descriptors and then cutting them back with variable selection techniques. Variable selection is an effective method to reduce the dimensionality but may discard some valuable information. This paper introduces Locally Linear Embedding ({LLE}), a local non-linear dimensionality reduction technique, that can statistically discover a low-dimensional representation of the chemical data. {LLE} is shown to create more stable representations than other non-linear dimensionality
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3103 reduction algorithms, and to be capable of capturing non-linearity in chemical data.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3104 topics={Bioinformatic},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3105 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3106
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3107 @TECHREPORT{lm-TR00,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3108 author = {Bengio, Yoshua and Ducharme, R{\'{e}}jean and Vincent, Pascal},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3109 title = {A Neural Probabilistic Language Model},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3110 number = {1178},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3111 year = {2000},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3112 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3113 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/TR1178.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3114 abstract = {A goal of statistical language modeling is to learn the joint probability function of sequences of words in a language. This is intrinsically difficult because of the curse of dimensionality: a word sequence on which the model will be tested is likely to be different from all the word sequences seen during training. Traditional but very successful approaches based on n-grams obtain generalization by concatenating very short overlapping sequences seen in the training set. We propose to fight the curse of dimensionality by learning a distributed representation for words which allows each training sentence to inform the model about an exponential number of semantically neighboring sentences. The model learns simultaneously (1) a distributed representation for each word along with (2) the probability function for word sequences, expressed in terms of these representations. Generalization is obtained because a sequence of words that has never been seen before gets high probability if it is made or words that are similar (in the sense of having a nearby representation) to words forming an already seen sentence. Training such large models (with millions of parameters) within a reasonable time is itself a significant challenge. We report on experiments using neural networks for the probability function, showing on two text corpora that the proposed approach very significantly improves on a state-of-the-art trigram model, and that the proposed approach allows to take advantage of much longer contexts.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3115 topics={Markov,Unsupervised,Language},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3116 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3117
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3118 @INPROCEEDINGS{Maillet+al-2009,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3119 author = {Maillet, Fran{\c c}ois and Eck, Douglas and Desjardins, Guillaume and Lamere, Paul},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3120 title = {Steerable Playlist Generation by Learning Song Similarity from Radio Station Playlists},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3121 booktitle = {Proceedings of the 10th International Conference on Music Information Retrieval},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3122 year = {2009},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3123 url = {http://www-etud.iro.umontreal.ca/~mailletf/papers/ismir09-playlist.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3124 abstract = {This paper presents an approach to generating steerable playlists. We ﬁrst demonstrate a method for learning song transition probabilities from audio features extracted from songs played in professional radio station playlists. We then show that by using this learnt similarity function as a prior, we are able to generate steerable playlists by choosing the next song to play not simply based on that prior, but on a tag cloud that the user is able to manipulate to express the high-level characteristics of the music he wishes Last.fm, to listen to.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3125 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3126
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3127 @INPROCEEDINGS{manzagol+bertinmahieux+eck:ismir2008,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3128 author = {Manzagol, Pierre-Antoine and Bertin-Mahieux, Thierry and Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3129 title = {On the Use of Sparse Time-Relative Auditory Codes for Music},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3130 booktitle = {{Proceedings of the 9th International Conference on Music Information Retrieval ({ISMIR} 2008)}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3131 year = {2008},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3132 abstract = {Many if not most audio features used in MIR research are inspired by work done in speech recognition and are variations on the spectrogram. Recently, much attention has been given to new representations of audio that are sparse and time-relative. These representations are efficient and able to avoid the time-frequency trade-off of a spectrogram. Yet little work with music streams has been conducted and these features remain mostly unused in the MIR community. In this paper we further explore the use of these features for musical signals. In particular, we investigate their use on realistic music examples (i.e. released commercial music) and their use as input features for supervised learning. Furthermore, we identify three specific issues related to these features which will need to be further addressed in order to obtain the full benefit for MIR applications.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3133 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3134 sourcetype={Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3135 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3136
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3137 @MASTERSTHESIS{Manzagol-Msc-2007,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3138 author = {Manzagol, Pierre-Antoine},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3139 key = {Algorithme d'apprentissage, méthode de second ordre, gradient naturel, approximation stochastique},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3140 title = {TONGA - Un algorithme de gradient naturel pour les probl{\`{e}}mes de grande taille},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3141 year = {2007},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3142 school = {Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3143 abstract = {Les syst{\`{e}}mes adaptatifs sont confront{\'{e}}s {\`{a}} des donn{\'{e}}es qui {\'{e}}voluent rapidement en quantit{\'{e}} et en complexit{\'{e}}. Les avanc{\'{e}}es mat{\'{e}}rielles de l'informatique ne susent pas {\`{a}} compenser cet essor. Une mise {\`{a}} l'{\'{e}}chelle des techniques d'apprentissage est n{\'{e}}cessaire. D'une part, les mod{\`{e}}les doivent gagner en capacit{\'{e}} de repr{\'{e}}sentation. De l'autre, les algorithmes d'apprentissage doivent devenir plus ecaces.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3144 Nos travaux se situent dans ce contexte des probl{\`{e}}mes de grande taille et portent sur l'am{\'{e}}lioration des algorithmes d'apprentissage. Deux {\'{e}}l{\'{e}}ments de r{\'{e}}ponse sont d{\'{e}}j{\`{a}} connus. Il s'agit des m{\'{e}}thodes de second ordre et de l'approximation stochastique. Or, les m{\'{e}}thodes de second ordre poss{\`{e}}dent des complexit{\'{e}}s en calculs et en m{\'{e}}moire qui sont prohibitives dans le cadre des probl{\`{e}}mes de grande taille. {\'{E}}galement, il est notoirement dicile de concilier ces m{\'{e}}thodes avec l'approximation stochastique. TONGA est un algorithme d'apprentissage con{\c c}u pour faire face {\`{a}} ces dicult{\'{e}}s. Il s'agit d'une implantation stochastique et adapt{\'{e}}e aux probl{\`{e}}mes de grande taille d'une m{\'{e}}thode de second ordre, le gradient naturel. Dans ce m{\'{e}}moire, nous examinons de pr{\`{e}}s ce nouvel algorithme d'apprentissage en le comparant sur plusieurs probl{\`{e}}mes au gradient stochastique, la technique d'optimisation commun{\'{e}}ment utilis{\'{e}}e dans le cadre des probl{\`{e}}mes de grande taille. Nos exp{\'{e}}riences montrent que TONGA est au moins tout aussi ecace que le gradient stochastique, ce qui est un accomplissement en soit. Dans certains cas, TONGA offre une convergence nettement sup{\'{e}}rieure {\`{a}} celle du gradient stochastique.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3145 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3146
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3147 @INPROCEEDINGS{matic-94,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3148 author = {Matic, N. and Henderson, Donnie and {LeCun}, Yann and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3149 title = {Pen-based visitor registration system (PENGUIN)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3150 booktitle = {Conference Record of the Twenty-Eighth Asilomar Conference on Signals, Systems and Computers},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3151 year = {1994},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3152 publisher = {IEEE},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3153 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/matic-94.tiff},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3154 abstract = {We describe a new electronic pen-based visitors registration system (PENGUIN) whose goal is to expand and modernize the visitor sign-in procedure at Bell Laboratories. The system uses a pen-interface (i.e. tablet-display) in what is essentially a form filling application. Our pen-interface is coupled with a powerful and accurate on-line handwriting recognition module. A database of AT&T employees (the visitors' hosts) and country names is used to check the recognition module outputs, in order to find the best match. The system provides assistance to the guard at one of the guard stations in routing visitors to their hosts. All the entered data are stored electronically. Initial testing shows that PENGUIN system performs reliably and with high accuracy. It retrieves the correct host name with 97\% accuracy and the correct visitors citizenship with 99\% accuracy. The system is robust and easy to use for both visitors and guards},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3155 topics={Speech},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3156 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3157
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3158 @UNPUBLISHED{mirex2005artist,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3159 author = {Bergstra, James and Casagrande, Norman and Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3160 title = {Artist Recognition: A Timbre- and Rhythm-Based Multiresolution Approach},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3161 year = {2005},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3162 note = {{MIREX} artist recognition contest},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3163 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3164 sourcetype={Other},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3165 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3166
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3167 @UNPUBLISHED{mirex2005genre,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3168 author = {Bergstra, James and Casagrande, Norman and Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3169 title = {Genre Classification: Timbre- and Rhythm-Based Multiresolution Audio Classification},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3170 year = {2005},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3171 note = {{MIREX} genre classification contest},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3172 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3173 sourcetype={Other},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3174 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3175
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3176 @UNPUBLISHED{mirex2005note,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3177 author = {Lacoste, Alexandre and Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3178 title = {Onset Detection with Artificial Neural Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3179 year = {2005},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3180 note = {{MIREX} note onset detection contest},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3181 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3182 sourcetype={Other},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3183 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3184
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3185 @UNPUBLISHED{mirex2005tempo,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3186 author = {Eck, Douglas and Casagrande, Norman},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3187 title = {A Tempo-Extraction Algorithm Using an Autocorrelation Phase Matrix and Shannon Entropy},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3188 year = {2005},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3189 note = {{MIREX} tempo extraction contest (www.music-ir.org/\-evaluation/\-mirex-results)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3190 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3191 sourcetype={Other},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3192 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3193
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3194 @INPROCEEDINGS{mitacs-insurance01,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3195 author = {Bengio, Yoshua and Chapados, Nicolas and Dugas, Charles and Ghosn, Joumana and Takeuchi, Ichiro and Vincent, Pascal},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3196 title = {High-Dimensional Data Inference for Automobile Insurance Premia Estimation},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3197 booktitle = {Presented at the 2001 MITACS Annual Meeting},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3198 year = {2001},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3199 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/mitacs_insurance.ps},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3200 topics={HighDimensional,Mining},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3201 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3202
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3203 @INPROCEEDINGS{Morin+al-2005,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3204 author = {Morin, Frederic and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3205 editor = {Cowell, Robert G. and Ghahramani, Zoubin},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3206 title = {Hierarchical Probabilistic Neural Network Language Model},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3207 booktitle = {Proceedings of the Tenth International Workshop on Artificial Intelligence and Statistics},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3208 year = {2005},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3209 pages = {246--252},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3210 publisher = {Society for Artificial Intelligence and Statistics},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3211 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/hierarchical-nnlm-aistats05.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3212 abstract = {In recent years, variants of a neural network architecture for statistical language modeling have been proposed and successfully applied, e.g. in the language modeling component of speech recognizers. The main advantage of these architectures is that they learn an embedding for words (or other symbols) in a continuous space that helps to smooth the language model and provide good generalization even when the number of training examples is insufficient. However, these models are extremely slow in comparison to the more commonly used n-gram models, both for training and recognition. As an alternative to an importance sampling method proposed to speed-up training, we introduce a hierarchical decomposition of the conditional probabilities that yields a speed-up of about 200 both during training and recognition. The hierarchical decomposition is a binary hierarchical clustering constrained by the prior knowledge extracted from the WordNet semantic hierarchy.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3213 topics={Language},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3214 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3215
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3216 @TECHREPORT{Nadeau-inference-TR99,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3217 author = {Nadeau, Claude and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3218 title = {Inference and the Generalization Error},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3219 number = {99s-45},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3220 year = {1999},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3221 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3222 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/techrep.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3223 abstract = {We perform a theoretical investigation of the variance of the cross-validation estimate of the generalization error that takes into account the variability due to the choice of training sets and test examples. This allows us to propose two new estimators of this variance. We show, via simulations, that these new statistics perform well relative to the statistics considered in (Dietterich, 1998). In particular, tests of hypothesis based on these dont tend to be too liberal like other tests currently available, and have good power.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3224 topics={Comparative},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3225 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3226
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3227 @INPROCEEDINGS{nadeau:2000:nips,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3228 author = {Nadeau, Claude and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3229 title = {Inference for the Generalization Error},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3230 year = {2000},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3231 pages = {307--313},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3232 crossref = {NIPS12-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3233 abstract = {In order to to compare learning algorithms, experimental results reported in the machine learning litterature often use statistical tests of significance. Unfortunately, most of these tests do not take into account the variability due to the choice of training set. We perform a theoretical investigation of the variance of the cross-validation estimate of the generalization error that takes into account the variability due to the choice of training sets. This allows us to propose two new ways to estimate this variance. We show, via simulations, that these new statistics perform well relative to the statistics considered by Dietterich (Dietterich, 1998).},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3234 topics={Comparative},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3235 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3236
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3237 @ARTICLE{nadeau:2001,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3238 author = {Nadeau, Claude and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3239 title = {Inference for the Generalization Error},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3240 journal = {Machine Learning},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3241 year = {2001},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3242 abstract = {In order to compare learning algorithms, experimental results reported in the machine learning literature often use statistical tests of significance to support the claim that a new learning algorithm generalizes better. Such tests should take into account the variability due to the choice of training set and not only that due to the test examples, as is often the case. This could lead to gross underestimation of the variance of the cross-validation estimator, and to the wrong conclusion that the new algorithm is significantly better when it is not. We perform a theoretical investigation of the variance of a cross-validation estimator of the generalization error that takes into account the variability due to the randomness of the training set as well as test examples. Our analysis shows that all the variance estimators that are based only on the results of the cross-validation experiment must be biased. This analysis allows us to propose new estimators of this variance. We show, via simulations, that tests of hypothesis about the generalization error using those new variance estimators have better properties than tests involving variance estimators currently in use and listed in (Dietterich, 1998). In particular, the new tests have correct size and good power. That is, the new tests do not reject the null hypothesis too often when the hypothesis is true, but they tend to frequently reject the null hypothesis when the latter is false.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3243 topics={Comparative},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3244 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3245
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3246 @ARTICLE{NC06,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3247 author = {Bengio, Yoshua and Monperrus, Martin and Larochelle, Hugo},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3248 title = {Nonlocal Estimation of Manifold Structure},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3249 journal = {Neural Computation},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3250 volume = {18},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3251 year = {2006},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3252 pages = {2509--2528},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3253 abstract = {We claim and present arguments to the effect that a large class of manifold
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3254 learning algorithms that are essentially local and can be framed as
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3255 kernel learning algorithms will suffer from the curse of dimensionality, at
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3256 the dimension of the true underlying manifold. This observation suggests
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3257 to explore non-local manifold learning algorithms which attempt to discover
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3258 shared structure in the tangent planes at different positions. A criterion for
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3259 such an algorithm is proposed and experiments estimating a tangent plane
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3260 prediction function are presented, showing its advantages with respect to
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3261 local manifold learning algorithms: it is able to generalize very far from
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3262 training data (on learning handwritten character image rotations), where a
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3263 local non-parametric method fails.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3264 topics={HighDimensional,Kernel,Unsupervised},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3265 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3266
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3267 @INPROCEEDINGS{NIPS1-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3268 editor = {Touretzky, D. S.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3269 title = {Advances in Neural Information Processing Systems 1 (NIPS'88)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3270 booktitle = {NIPS 1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3271 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3272 publisher = {Morgan Kaufmann}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3273 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3274
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3275
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3276 @INPROCEEDINGS{NIPS10-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3277 editor = {Jordan, M.I. and Kearns, M.J. and Solla, S.A.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3278 title = {Advances in Neural Information Processing Systems 10 (NIPS'97)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3279 booktitle = {NIPS 10},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3280 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3281 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3282 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3283
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3284
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3285 @INPROCEEDINGS{NIPS11,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3286 editor = {Kearns, M.J. and Solla, S.A.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3287 title = {Advances in Neural Information Processing Systems 11 (NIPS'98)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3288 booktitle = {Advances in Neural Information Processing Systems 11 (NIPS'98)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3289 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3290 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3291 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3292
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3293 @INPROCEEDINGS{NIPS11-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3294 editor = {Kearns, M.J. and Solla, S.A.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3295 title = {Advances in Neural Information Processing Systems 11 (NIPS'98)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3296 booktitle = {NIPS 11},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3297 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3298 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3299 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3300
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3301
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3302 @INPROCEEDINGS{NIPS12-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3303 editor = {Solla, S.A. and Leen, T. K.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3304 title = {Advances in Neural Information Processing Systems 12 (NIPS'99)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3305 booktitle = {NIPS 12},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3306 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3307 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3308 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3309
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3310
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3311 @INPROCEEDINGS{NIPS13-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3312 editor = {Leen, T. K. and Dietterich, T.G.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3313 title = {Advances in Neural Information Processing Systems 13 (NIPS'00)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3314 booktitle = {NIPS 13},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3315 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3316 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3317 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3318
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3319
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3320 @INPROCEEDINGS{NIPS14,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3321 editor = {Dietterich, T.G. and Becker, S. and Ghahramani, Zoubin},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3322 title = {Advances in Neural Information Processing Systems 14 (NIPS'01)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3323 booktitle = {Advances in Neural Information Processing Systems 14 (NIPS'01)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3324 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3325 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3326 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3327
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3328 @INPROCEEDINGS{NIPS14-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3329 editor = {Dietterich, T.G. and Becker, S. and Ghahramani, Zoubin},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3330 title = {Advances in Neural Information Processing Systems 14 (NIPS'01)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3331 booktitle = {NIPS 14},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3332 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3333 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3334 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3335
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3336
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3337 @INPROCEEDINGS{NIPS15-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3338 editor = {Becker, S. and Thrun, Sebastian},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3339 title = {Advances in Neural Information Processing Systems 15 (NIPS'02)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3340 booktitle = {NIPS 15},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3341 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3342 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3343 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3344
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3345
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3346 @INPROCEEDINGS{NIPS16-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3347 editor = {Becker, S. and Saul, L. and {Sch{\"{o}}lkopf}, Bernhard},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3348 title = {Advances in Neural Information Processing Systems 16 (NIPS'03)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3349 booktitle = {NIPS 16},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3350 year = {-1}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3351 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3352
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3353
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3354 @INPROCEEDINGS{NIPS17-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3355 editor = {Saul, Lawrence K. and Weiss, Yair and Bottou, {L{\'{e}}on}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3356 title = {Advances in Neural Information Processing Systems 17 (NIPS'04)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3357 booktitle = {NIPS 17},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3358 year = {-1}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3359 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3360
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3361
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3362 @INPROCEEDINGS{NIPS18-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3363 editor = {Weiss, Yair and {Sch{\"{o}}lkopf}, Bernhard and Platt, John},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3364 title = {Advances in Neural Information Processing Systems 18 (NIPS'05)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3365 booktitle = {NIPS 18},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3366 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3367 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3368 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3369
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3370
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3371 @INPROCEEDINGS{NIPS19-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3372 editor = {{Sch{\"{o}}lkopf}, Bernhard and Platt, John and Hoffman, Thomas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3373 title = {Advances in Neural Information Processing Systems 19 (NIPS'06)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3374 booktitle = {NIPS 19},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3375 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3376 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3377 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3378
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3379
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3380 @INPROCEEDINGS{NIPS2-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3381 editor = {Touretzky, D. S.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3382 title = {Advances in Neural Information Processing Systems 2 (NIPS'89)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3383 booktitle = {NIPS 2},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3384 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3385 publisher = {Morgan Kaufmann}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3386 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3387
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3388
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3389 @INPROCEEDINGS{NIPS20-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3390 editor = {Platt, John and Koller, D. and Singer, Yoram and Roweis, S.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3391 title = {Advances in Neural Information Processing Systems 20 (NIPS'07)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3392 booktitle = {NIPS 20},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3393 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3394 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3395 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3396
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3397
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3398 @INPROCEEDINGS{NIPS2003_AA65,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3399 author = {Bengio, Yoshua and Grandvalet, Yves},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3400 keywords = {cross validation, error bars, generalization error inference, k-fold cross-validation, model selection, statistical comparison of algorithms, variance estimate},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3401 title = {No Unbiased Estimator of the Variance of K-Fold Cross-Validation},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3402 year = {2004},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3403 publisher = {MIT Press},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3404 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/var-kfold-part1-nips.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3405 crossref = {NIPS16},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3406 abstract = {Most machine learning researchers perform quantitative experiments to estimate generalization error and compare algorithm performances. In order to draw statistically convincing conclusions, it is important to estimate the uncertainty of such estimates. This paper studies the estimation of uncertainty around the K-fold cross-validation estimator. The main theorem shows that there exists no universal unbiased estimator of the variance of K-fold cross-validation. An analysis based on the eigendecomposition of the covariance matrix of errors helps to better understand the nature of the problem and shows that naive estimators may grossly underestimate variance, as confirmed by numerical experiments.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3407 topics={Comparative},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3408 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3409
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3410 @INCOLLECTION{NIPS2005_424,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3411 author = {Bengio, Yoshua and Delalleau, Olivier and Le Roux, Nicolas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3412 title = {The Curse of Highly Variable Functions for Local Kernel Machines},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3413 year = {2006},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3414 pages = {107--114},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3415 crossref = {NIPS18-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3416 abstract = {We present a series of theoretical arguments supporting the claim that a
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3417 large class of modern learning algorithms that rely solely on the smoothness
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3418 prior with similarity between examples expressed with a local
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3419 kernel are sensitive to the curse of dimensionality, or more precisely
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3420 to the variability of the target. Our discussion covers supervised, semisupervised
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3421 and unsupervised learning algorithms. These algorithms are
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3422 found to be local in the sense that crucial properties of the learned function
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3423 at x depend mostly on the neighbors of x in the training set. This
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3424 makes them sensitive to the curse of dimensionality, well studied for
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3425 classical non-parametric statistical learning. We show in the case of the
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3426 Gaussian kernel that when the function to be learned has many variations,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3427 these algorithms require a number of training examples proportional to
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3428 the number of variations, which could be large even though there may exist
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3429 short descriptions of the target function, i.e. their Kolmogorov complexity
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3430 may be low. This suggests that there exist non-local learning
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3431 algorithms that at least have the potential to learn about such structured
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3432 but apparently complex functions (because locally they have many variations),
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3433 while not using very specific prior domain knowledge.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3434 topics={HighDimensional,Kernel,Unsupervised},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3435 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3436
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3437 @INPROCEEDINGS{NIPS2005_456,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3438 author = {K{\'{e}}gl, Bal{\'{a}}zs and Wang, Ligen},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3439 title = {Boosting on Manifolds: Adaptive Regularization of Base Classifiers},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3440 year = {2005},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3441 pages = {665--672},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3442 crossref = {NIPS17-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3443 abstract = {In this paper we propose to combine two powerful ideas, boosting and manifold learning. On the one hand, we improve ADABOOST by incorporating knowledge on the structure of the data into base classifier design and selection. On the other hand, we use ADABOOSTs efficient learning mechanism to significantly improve supervised and semi-supervised algorithms proposed in the context of manifold learning. Beside the specific manifold-based penalization, the resulting algorithm also accommodates the boosting of a large family of regularized learning algorithms.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3444 topics={Boosting},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3445 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3446
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3447 @INCOLLECTION{NIPS2005_519,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3448 author = {Grandvalet, Yves and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3449 title = {Semi-supervised Learning by Entropy Minimization},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3450 year = {2005},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3451 pages = {529--236},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3452 crossref = {NIPS17-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3453 abstract = {We consider the semi-supervised learning problem, where a decision rule is to be learned from labeled and unlabeled data. In this framework, we motivate minimum entropy regularization, which enables to incorporate unlabeled data in the standard supervised learning. Our approach includes other approaches to the semi-supervised problem as particular or limiting cases. A series of experiments illustrates that the proposed solution benefits from unlabeled data. The method challenges mixture models when the data are sampled from the distribution class spanned by the generative model. The performances are definitely in favor of minimum entropy regularization when generative models are misspecified, and the weighting of unlabeled data provides robustness to the violation of the cluster assumption. Finally, we also illustrate that the method can also be far superior to manifold learning in high dimension spaces.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3454 topics={Unsupervised},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3455 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3456
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3457 @INPROCEEDINGS{NIPS2005_539,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3458 author = {Bengio, Yoshua and Larochelle, Hugo and Vincent, Pascal},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3459 title = {Non-Local Manifold Parzen Windows},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3460 year = {2006},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3461 crossref = {NIPS18-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3462 abstract = {To escape from the curse of dimensionality, we claim that one can learn
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3463 non-local functions, in the sense that the value and shape of the learned
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3464 function at x must be inferred using examples that may be far from x.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3465 With this objective, we present a non-local non-parametric density estimator.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3466 It builds upon previously proposed Gaussian mixture models with
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3467 regularized covariance matrices to take into account the local shape of
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3468 the manifold. It also builds upon recent work on non-local estimators of
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3469 the tangent plane of a manifold, which are able to generalize in places
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3470 with little training data, unlike traditional, local, non-parametric models.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3471 topics={HighDimensional,Kernel,Unsupervised},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3472 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3473
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3474 @INPROCEEDINGS{NIPS2005_583,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3475 author = {Bengio, Yoshua and Le Roux, Nicolas and Vincent, Pascal and Delalleau, Olivier and Marcotte, Patrice},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3476 title = {Convex Neural Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3477 year = {2006},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3478 pages = {123--130},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3479 crossref = {NIPS18-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3480 abstract = {Convexity has recently received a lot of attention in the machine learning
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3481 community, and the lack of convexity has been seen as a major disadvantage
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3482 of many learning algorithms, such as multi-layer artificial neural
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3483 networks. We show that training multi-layer neural networks in which the
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3484 number of hidden units is learned can be viewed as a convex optimization
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3485 problem. This problem involves an infinite number of variables, but can be
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3486 solved by incrementally inserting a hidden unit at a time, each time finding
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3487 a linear classifier that minimizes a weighted sum of errors.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3488 topics={Boosting},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3489 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3490
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3491 @INPROCEEDINGS{NIPS2005_663,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3492 author = {Rivest, Fran{\c c}ois and Bengio, Yoshua and Kalaska, John},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3493 title = {Brain Inspired Reinforcement Learning},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3494 year = {2005},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3495 pages = {1129--1136},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3496 crossref = {NIPS17-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3497 abstract = {Successful application of reinforcement learning algorithms often involves considerable hand-crafting of the necessary non-linear features to reduce the complexity of the value functions and hence to promote convergence of the algorithm. In contrast, the human brain readily and autonomously finds the complex features when provided with sufficient training. Recent work in machine learning and neurophysiology has demonstrated the role of the basal ganglia and the frontal cortex in mammalian reinforcement learning. This paper develops and explores new reinforcement learning algorithms inspired by neurological evidence that provides potential new approaches to the feature construction problem. The algorithms are compared and evaluated on the Acrobot task.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3498 topics={BioRules},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3499 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3500
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3501 @INCOLLECTION{NIPS2005_691,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3502 author = {Bengio, Yoshua and Monperrus, Martin},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3503 title = {Non-Local Manifold Tangent Learning},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3504 year = {2005},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3505 pages = {129--136},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3506 crossref = {NIPS17-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3507 abstract = {We claim and present arguments to the effect that a large class of manifold learning algorithms that are essentially local and can be framed as kernel learning algorithms will suffer from the curse of dimensionality, at the dimension of the true underlying manifold. This observation suggests to explore non-local manifold learning algorithms which attempt to discover shared structure in the tangent planes at different positions. A criterion for such an algorithm is proposed and experiments estimating a tangent plane prediction function are presented, showing its advantages with respect to local manifold learning algorithms: it is able to generalize very far from training data (on learning handwritten character image rotations), where a local non-parametric method fails.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3508 topics={HighDimensional,Unsupervised},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3509 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3510
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3511 @INPROCEEDINGS{NIPS2005_874,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3512 author = {K{\'{e}}gl, Bal{\'{a}}zs},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3513 title = {Generalization Error and Algorithmic Convergence of Median Boosting},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3514 year = {2005},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3515 pages = {657--664},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3516 crossref = {NIPS17-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3517 abstract = {We have recently proposed an extension of ADABOOST to regression that uses the median of the base regressors as the final regressor. In this paper we extend theoretical results obtained for ADABOOST to median boosting and to its localized variant. First, we extend recent results on efficient margin maximizing to show that the algorithm can converge to the maximum achievable margin within a preset precision in a finite number of steps. Then we provide confidence-interval-type bounds on the generalization error.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3518 topics={Boosting},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3519 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3520
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3521 @INPROCEEDINGS{NIPS2007-56,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3522 author = {Le Roux, Nicolas and Manzagol, Pierre-Antoine and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3523 title = {Topmoumoute online natural gradient algorithm},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3524 year = {2008},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3525 crossref = {NIPS20-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3526 abstract = {Guided by the goal of obtaining an optimization algorithm that is both fast and yielding good generalization, we study the descent direction maximizing the decrease in generalization error or the probability of not increasing generalization error. The surprising result is that from both the Bayesian and frequentist perspectives this can yield the natural gradient direction. Although that direction can be very expensive to compute we develop an efficient, general, online approximation to the natural gradient descent which is suited to large scale problems. We report experimental results showing much faster convergence in computation time and in number of iterations with TONGA (Topmoumoute Online natural Gradient Algorithm) than with stochastic gradient descent, even on very large datasets.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3527 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3528
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3529 @INPROCEEDINGS{NIPS2007-812,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3530 author = {Chapados, Nicolas and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3531 title = {Augmented Functional Time Series Representation and Forecasting with Gaussian Processes},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3532 year = {2008},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3533 pages = {265--272},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3534 crossref = {NIPS20-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3535 abstract = {We introduce a functional representation of time series which allows forecasts to be performed over an unspecified horizon with progressively-revealed information sets. By virtue of using Gaussian processes, a complete covariance matrix between forecasts at several time-steps is available. This information is put to use in an application to actively trade price spreads between commodity futures contracts. The approach delivers impressive out-of-sample risk-adjusted returns after transaction costs on a portfolio of 30 spreads.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3536 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3537
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3538 @INPROCEEDINGS{NIPS2007-925,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3539 author = {Le Roux, Nicolas and Bengio, Yoshua and Lamblin, Pascal and Joliveau, Marc and K{\'{e}}gl, Bal{\'{a}}zs},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3540 title = {Learning the 2-D Topology of Images},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3541 year = {2008},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3542 pages = {841--848},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3543 crossref = {NIPS20-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3544 abstract = {We study the following question: is the two-dimensional structure of images a very strong prior or is it something that can be learned with a few examples of natural images? If someone gave us a learning task involving images for which the two-dimensional topology of pixels was not known, could we discover it automatically and exploit it? For example suppose that the pixels had been permuted in a fixed but unknown way, could we recover the relative two-dimensional location of pixels on images? The surprising result presented here is that not only the answer is yes but that about as few as a thousand images are enough to approximately recover the relative locations of about a thousand pixels. This is achieved using a manifold learning algorithm applied to pixels associated with a measure of distributional similarity between pixel intensities. We compare different topologyextraction approaches and show how having the two-dimensional topology can be exploited.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3545 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3546
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3547 @INPROCEEDINGS{NIPS21,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3548 editor = {Koller, D. and Schuurmans, Dale and Bengio, Yoshua and Bottou, {L{\'{e}}on}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3549 title = {Advances in Neural Information Processing Systems 21 (NIPS'08)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3550 booktitle = {Advances in Neural Information Processing Systems 21 (NIPS'08)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3551 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3552 publisher = {Nips Foundation (http://books.nips.cc)}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3553 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3554
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3555 @INPROCEEDINGS{NIPS21-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3556 editor = {Koller, D. and Schuurmans, Dale and Bengio, Yoshua and Bottou, {L{\'{e}}on}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3557 title = {Advances in Neural Information Processing Systems 21 (NIPS'08)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3558 booktitle = {NIPS 21},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3559 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3560 publisher = {Nips Foundation (http://books.nips.cc)}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3561 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3562
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3563
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3564 @INPROCEEDINGS{NIPS22-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3565 editor = {Bengio, Yoshua and Schuurmans, Dale and Williams, Christopher and Lafferty, John and Culotta, Aron},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3566 title = {Advances in Neural Information Processing Systems 22 (NIPS'09)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3567 booktitle = {NIPS 22},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3568 year = {-1}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3569 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3570
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3571
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3572 @INPROCEEDINGS{NIPS3,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3573 editor = {Lipmann, R. P. and Moody, J. E. and Touretzky, D. S.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3574 title = {Advances in Neural Information Processing Systems 3 (NIPS'90)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3575 booktitle = {Advances in Neural Information Processing Systems 3 (NIPS'90)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3576 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3577 publisher = {Morgan Kaufmann}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3578 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3579
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3580 @INPROCEEDINGS{NIPS3-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3581 editor = {Lipmann, R. P. and Moody, J. E. and Touretzky, D. S.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3582 title = {Advances in Neural Information Processing Systems 3 (NIPS'90)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3583 booktitle = {NIPS 3},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3584 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3585 publisher = {Morgan Kaufmann}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3586 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3587
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3588
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3589 @INPROCEEDINGS{NIPS4-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3590 editor = {Moody, J. E. and Hanson, S. J. and Lipmann, R. P.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3591 title = {Advances in Neural Information Processing Systems 4 (NIPS'91)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3592 booktitle = {NIPS 4},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3593 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3594 publisher = {Morgan Kaufmann}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3595 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3596
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3597
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3598 @INPROCEEDINGS{NIPS5,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3599 editor = {Giles, C.L. and Hanson, S. J. and Cowan, J. D.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3600 title = {Advances in Neural Information Processing Systems 5 (NIPS'92)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3601 booktitle = {Advances in Neural Information Processing Systems 5 (NIPS'92)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3602 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3603 publisher = {Morgan Kaufmann}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3604 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3605
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3606 @INPROCEEDINGS{NIPS5-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3607 editor = {Giles, C.L. and Hanson, S. J. and Cowan, J. D.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3608 title = {Advances in Neural Information Processing Systems 5 (NIPS'92)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3609 booktitle = {NIPS 5},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3610 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3611 publisher = {Morgan Kaufmann}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3612 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3613
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3614
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3615 @INPROCEEDINGS{NIPS6-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3616 editor = {Cowan, J. D. and Tesauro, G. and Alspector, J.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3617 title = {Advances in Neural Information Processing Systems 6 (NIPS'93)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3618 booktitle = {NIPS 6},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3619 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3620 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3621 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3622
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3623
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3624 @INPROCEEDINGS{NIPS7-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3625 editor = {Tesauro, G. and Touretzky, D. S. and Leen, T. K.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3626 title = {Advances in Neural Information Processing Systems 7 (NIPS'94)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3627 booktitle = {NIPS 7},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3628 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3629 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3630 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3631
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3632
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3633 @INPROCEEDINGS{NIPS8-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3634 editor = {Touretzky, D. S. and Mozer, M. and Hasselmo, M.E.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3635 title = {Advances in Neural Information Processing Systems 8 (NIPS'95)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3636 booktitle = {NIPS 8},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3637 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3638 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3639 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3640
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3641
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3642 @INPROCEEDINGS{NIPS9-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3643 editor = {Mozer, M. and Jordan, M.I. and Petsche, T.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3644 title = {Advances in Neural Information Processing Systems 9 (NIPS'96)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3645 booktitle = {NIPS 9},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3646 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3647 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3648 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3649
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3650
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3651 @INPROCEEDINGS{nnlm:2001:nips,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3652 author = {Bengio, Yoshua and Ducharme, R{\'{e}}jean and Vincent, Pascal},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3653 title = {A Neural Probabilistic Language Model},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3654 year = {2001},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3655 crossref = {NIPS13-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3656 abstract = {A goal of statistical language modeling is to learn the joint probability function of sequences of words. This is intrinsically difficult because of the curse of dimensionality: we propose to fight it with its own weapons. In the proposed approach one learns simultaneously (1) a distributed representation for each word (i.e. a similarity between words) along with (2) the probability function for word sequences, expressed with these representations. Generalization is obtained because a sequence of words that
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3657 has never been seen before gets high probability if it is made of words that are similar to words forming an already seen sentence. We report on experiments using neural networks for the probability function, showing on two text corpora that the proposed approach very significantly improves on a state-of-the-art trigram model.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3658 topics={Markov,Unsupervised,Language},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3659 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3660
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3661 @INPROCEEDINGS{nsvn:2000:ijcnn,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3662 author = {Vincent, Pascal and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3663 title = {A Neural Support Vector Network Architecture with Adaptive Kernels},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3664 booktitle = {International Joint Conference on Neural Networks 2000},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3665 volume = {V},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3666 year = {2000},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3667 pages = {187--192},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3668 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/nsvn.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3669 abstract = {In the Support Vector Machines ({SVM}) framework, the positive-definite kernel can be seen as representing a fixed similarity measure between two patterns, and a discriminant function is obtained by taking a linear combination of the kernels computed at training examples called support vectors. Here we investigate learning architectures in which the kernel functions can be replaced by more general similarity measures that can have arbitrary internal parameters. The training criterion used in {SVM}s is not appropriate for this purpose so we adopt the simple criterion that is generally used when training neural networks for classification tasks. Several experiments are performed which show that such Neural Support Vector Networks perform similarly to {SVM}s while requiring significantly fewer support vectors, even when the similarity measure has no internal parameters.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3670 topics={Kernel},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3671 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3672
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3673 @INPROCEEDINGS{Ouimet+al-2005,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3674 author = {Ouimet, Marie and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3675 editor = {Cowell, Robert G. and Ghahramani, Zoubin},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3676 title = {Greedy Spectral Embedding},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3677 booktitle = {Proceedings of the Tenth International Workshop on Artificial Intelligence and Statistics},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3678 year = {2005},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3679 pages = {253--260},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3680 publisher = {Society for Artificial Intelligence and Statistics},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3681 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/greedy-kernel-aistats05.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3682 abstract = {Spectral dimensionality reduction methods and spectral clustering methods require computation of the principal eigenvectors of an n X n matrix where n is the number of examples. Following up on previously proposed techniques to speed-up kernel methods by focusing on a subset of m examples, we study a greedy selection procedure for this subset, based on the feature space distance between a candidate example and the span of the previously chosen ones. In the case of kernel {PCA} or spectral clustering this reduces computation to O(m^2 n). For the same computational complexity, we can also compute the feature space projection of the non-selected examples on the subspace spanned by the selected examples, to estimate the embedding function based on all the data, which yields considerably better estimation of the embedding function. This algorithm can be formulated in an online setting and we can bound the error on the approximation of the Gram matrix.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3683 topics={HighDimensional,kenel},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3684 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3685
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3686 @MASTERSTHESIS{Ouimet-Msc-2004,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3687 author = {Ouimet, Marie},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3688 keywords = {algorithmes voraces., apprentissage non-supervis{\'{e}}, m{\'{e}}thodes spectrales, noyaux, r{\'{e}}duction de dimensionnalit{\'{e}}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3689 title = {R{\'{e}}duction de dimensionnalit{\'{e}} non lin{\'{e}}aire et vorace},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3690 year = {2004},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3691 school = {Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3692 abstract = {Les m{\'{e}}thodes spectrales de r{\'{e}}duction de dimensionnalit{\'{e}} et les m{\'{e}}thodes de segmentation spectrale exigent le calcul des vecteurs propres principaux d'une matrice de taille n x n o{\`{u}} n est le nombre d'exemples. Des techniques ont {\'{e}}t{\'{e}} propos{\'{e}}es dans la litt{\'{e}}rature pour acc{\'{e}}l{\'{e}}rer les m{\'{e}}thodes {\`{a}} noyau en se concentrant sur un sous-ensemble de m exemples. Nous proposons une proc{\'{e}}dure vorace pour la s{\'{e}}lection de ce sous-ensemble, qui est bas{\'{e}}e sur la distance dans l'espace des caract{\`{e}}ristiques entre un exemple candidat et le sous-espace g{\'{e}}n{\'{e}}r{\'{e}} par les exemples pr{\'{e}}c{\'{e}}demment choisis. Dans le cas de l'ACP {\`{a}} noyau ou de la segmentation spectrale, nous obtenons un algorithme en O(mmn), o{\`{u}} m << n, qui, contrairement aux techniques pr{\'{e}}c{\'{e}}demment propos{\'{e}}es, peut se formuler de fa{\c c}on en-ligne. Pour la m{\^{e}}me complexit{\'{e}} en temps, nous pouvons {\'{e}}galement calculer la projection des exemples non choisis sur le sous-espace engendr{\'{e}} par les exemples choisis dans l'espace des caract{\'{e}}ristiques. En repr{\'{e}}sentant ainsi les exemples par leur projection nous obtenons une approximation de plus faible rang de la matrice de Gram sur toutes les donn{\'{e}}es. Nous pouvons {\'{e}}galement borner l'erreur correspondant {\`{a}} cette approximation de la matrice de Gram.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3693 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3694
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3695 @ARTICLE{paiement+bengio+eck:aij,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3696 author = {Paiement, Jean-Fran{\c c}ois and Bengio, Samy and Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3697 title = {Probabilistic Models for Melodic Prediction},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3698 journal = {Artificial Intelligence Journal},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3699 volume = {173},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3700 year = {2009},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3701 pages = {1266-1274},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3702 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3703 sourcetype={Journal},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3704 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3705
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3706 @INPROCEEDINGS{paiement+eck+bengio+barber:icml2005,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3707 author = {Paiement, Jean-Fran{\c c}ois and Eck, Douglas and Bengio, Samy and Barber, D.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3708 title = {A graphical model for chord progressions embedded in a psychoacoustic space},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3709 year = {2005},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3710 pages = {641--648},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3711 publisher = {ACM Press},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3712 crossref = {ICML05},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3713 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3714 sourcetype={Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3715 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3716
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3717 @INPROCEEDINGS{paiement+eck+bengio:ccai2006,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3718 author = {Paiement, Jean-Fran{\c c}ois and Eck, Douglas and Bengio, Samy},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3719 editor = {Lamontagne, Luc and Marchand, Mario},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3720 title = {Probabilistic Melodic Harmonization},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3721 booktitle = {Canadian Conference on AI},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3722 series = {Lecture Notes in Computer Science},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3723 volume = {4013},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3724 year = {2006},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3725 pages = {218-229},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3726 publisher = {Springer},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3727 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3728 sourcetype={Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3729 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3730
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3731 @INPROCEEDINGS{paiement+eck+bengio:ismir2005,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3732 author = {Paiement, Jean-Fran{\c c}ois and Eck, Douglas and Bengio, Samy},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3733 title = {A Probabilistic Model for Chord Progressions},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3734 booktitle = {{Proceedings of the 6th International Conference on Music Information Retrieval ({ISMIR} 2005)}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3735 year = {2005},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3736 pages = {312-319},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3737 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3738 sourcetype={Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3739 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3740
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3741 @INPROCEEDINGS{paiement+grandvalet+bengio+eck:icml2008,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3742 author = {Paiement, Jean-Fran{\c c}ois and Grandvalet, Yves and Bengio, Samy and Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3743 title = {A generative model for rhythms},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3744 year = {2008},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3745 pages = {},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3746 crossref = {ICML06-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3747 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3748 sourcetype={Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3749 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3750
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3751 @UNPUBLISHED{paiement+grandvalet+bengio+eck:nipsworkshop2007,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3752 author = {Paiement, Jean-Fran{\c c}ois and Grandvalet, Yves and Bengio, Samy and Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3753 title = {A generative model for rhythms},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3754 year = {2007},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3755 note = {NIPS 2007 Workshop on Music, Brain and Cognition},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3756 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3757 sourcetype={Workshop},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3758 optkey={""},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3759 optmonth={""},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3760 optannote={""},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3761 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3762
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3763 @MASTERSTHESIS{Paiement-Msc-2003,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3764 author = {Paiement, Jean-Fran{\c c}ois},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3765 keywords = {algorithmes, apprentissage, apprentissage non supervis{\'{e}}, forage de donn{\'{e}}es, noyaux, r{\'{e}}duction de dimensions, statistique, Statistiques},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3766 title = {G{\'{e}}n{\'{e}}ralisation d'algorithmes de r{\'{e}}duction de dimension},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3767 year = {2003},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3768 school = {Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3769 abstract = {On pr{\'{e}}sente tout d'abord la notion de vari{\'{e}}t{\'{e}} comme r{\'{e}}gion de faible dimension contenant des observations situ{\'{e}}es dans un espace de haute dimension. Cette d{\'{e}}finition justifie l'{\'{e}}laboration d'algorithmes permettant d'exprimer les donn{\'{e}}es dans un syst{\`{e}}me de coordonn{\'{e}}es de dimensions {\'{e}}gale {\`{a}} celle de la vari{\'{e}}t{\'{e}} sur laquelle les donn{\'{e}}es sont approximativement situ{\'{e}}es.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3770 La notion de noyau comme mesure de similarit{\'{e}} est par la suite formalis{\'{e}}e. On constate que l'application d'un noyau {\`{a}} deux observations correspond {\`{a}} l'{\'{e}}valuation d'un produit scalaire dans un espace de Hilbert appel{\'{e}} espace de caract{\'{e}}ristiques.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3771 Une m{\'{e}}thode de r{\'{e}}duction de dimension lin{\'{e}}raire est expos{\'{e}}e ainsi que ces limites. Des algorithmes non lin{\'{e}}raires de r{\'{e}}duction de dimension et de segmentation permettent de s'affranchir de ces limites. Ces derniers ne fournissent cependant pas d'extension directe {\`{a}} des points hors {\'{e}}chantillon.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3772 L'{\'{e}}tape fondamentale au sein des algorithmes pr{\'{e}}sent{\'{e}}s est la solution d'un syst{\`{e}}me de vecteurs propres d'une matrice sym{\'{e}}trique cr{\'{e}}{\'{e}}e {\`{a}} partir d'un noyau d{\'{e}}pendant des donn{\'{e}}es. On con{\c c}oit cd probl{\`{e}}me comme le fait de trouver les fonctions propres d'un op{\'{e}}rateur lin{\'{e}}aire d{\'{e}}fini {\`{a}} partir du m{\^{e}}me noyau. On utilise alors la formulation de Nystr{\"{o}}m, pr{\'{e}}sente dans l'espace en composantes principales {\`{a}} noyaux, afin de r{\'{e}}duire la dimension des points hors {\'{e}}chantillon sur la vase des plongements obtenus {\`{a}} l'aide des algorithmes d{\'{e}}j{\`{a}} mentionn{\'{e}}s.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3773 La qualit{\'{e}} de la projection g{\'{e}}n{\'{e}}r{\'{e}}e est compar{\'{e}}e {\`{a}} la perturbation intrins{\`{e}}que des algorithmes si on substitue certaine observations par d'autres tir{\'{e}}es de la m{\^{e}}me distribution.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3774 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3775
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3776 @ARTICLE{perez+gers+schmidhuber+eck:2002,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3777 author = {Perez-Ortiz, J. A. and Gers, F. A. and Eck, Douglas and Schmidhuber, Juergen},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3778 title = {{K}alman filters improve {LSTM} network performance in problems unsolvable by traditional recurrent nets},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3779 journal = {Neural Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3780 volume = {16},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3781 number = {2},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3782 year = {2003},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3783 abstract = {The Long Short-Term Memory ({LSTM}) network trained by gradient descent solves difficult problems which traditional recurrent neural networks in general cannot. We have recently observed that the decoupled extended Kalman filter training algorithm allows for even better performance, reducing significantly the number of training steps when compared to the original gradient descent training algorithm. In this paper we present a set of experiments which are unsolvable by classical recurrent networks but which are solved elegantly and robustly and quickly by {LSTM} combined with Kalman filters.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3784 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3785 sourcetype={Journal},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3786 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3787
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3788 @ARTICLE{perez+gers+schmidhuber+eck:2003,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3789 author = {Perez-Ortiz, J. A. and Gers, F. A. and Eck, Douglas and Schmidhuber, Juergen},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3790 title = {{K}alman filters improve {LSTM} network performance in problems unsolvable by traditional recurrent nets},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3791 journal = {Neural Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3792 volume = {16},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3793 number = {2},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3794 year = {2003},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3795 pages = {241--250},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3796 abstract = {The Long Short-Term Memory ({LSTM}) network trained by gradient descent solves difficult problems which traditional recurrent neural networks in general cannot. We have recently observed that the decoupled extended Kalman filter training algorithm allows for even better performance, reducing significantly the number of training steps when compared to the original gradient descent training algorithm. In this paper we present a set of experiments which are unsolvable by classical recurrent networks but which are solved elegantly and robustly and quickly by {LSTM} combined with Kalman filters.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3797 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3798 sourcetype={Journal},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3799 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3800
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3801 @INPROCEEDINGS{perez+schmidhuber+gers+eck:icannB2002,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3802 author = {Perez-Ortiz, J. A. and Schmidhuber, Juergen and Gers, F. A. and Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3803 editor = {Dorronsoro, J.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3804 title = {Improving Long-Term Online Prediction with {Decoupled Extended Kalman Filters}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3805 booktitle = {{Artificial Neural Networks -- ICANN 2002 (Proceedings)}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3806 year = {2002},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3807 pages = {1055--1060},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3808 publisher = {Springer},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3809 abstract = {Long Short-Term Memory ({LSTM}) recurrent neural networks ({RNN}s) outperform traditional {RNN}s when dealing with sequences involving not only short-term but also long-term dependencies. The decoupled extended Kalman filter learning algorithm ({DEKF}) works well in online environments and reduces significantly the number of training steps when compared to the standard gradient-descent algorithms. Previous work on {LSTM}, however, has always used a form of gradient descent and has not focused on true online situations. Here we combine {LSTM} with {DEKF} and show that this new hybrid improves upon the original learning algorithm when applied to online processing.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3810 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3811 sourcetype={Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3812 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3813
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3814 @TECHREPORT{Pigeon-Bengio-96-aH-TR,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3815 author = {Pigeon, Steven and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3816 title = {A Memory-Efficient Huffman Adaptive Coding Algorithm for Very Large Sets of Symbols},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3817 number = {\#1081},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3818 year = {1997},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3819 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3820 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/HuffAdapt.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3821 abstract = {The problem of computing the minimum redundancy codes as we observe symbols one by one has received a lot of attention. However, existing algorithm implicitly assumes that either we have a small alphabet quite typically 256 symbols or that we have an arbitrary amount of memory at our disposal for the creation of the tree. In real life applications one may need to encode symbols coming from a much larger alphabet, for e.g. coding integers. We now have to deal not with hundreds of symbols but possibly with millions of symbols. While other algorithms use a space proportional to the number of observed symbol, we here propose one that uses space proportional to the number of frequency classes, which is, quite interestingly, always smaller or equal to the number of observed symbols.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3822 topics={Compression},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3823 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3824
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3825 @INPROCEEDINGS{Pigeon-dcc98,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3826 author = {Pigeon, Steven and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3827 editor = {Society, {IEEE} Computer},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3828 title = {A Memory-Efficient Adaptive Huffman Coding Algorithm for Very Large Sets of Symbols},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3829 booktitle = {Data Compression Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3830 year = {1998},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3831 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/dcc98.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3832 abstract = {The problem of computing the minimum redundancy codes as we observe symbols one by one has received a lot of attention. However, existing algorithms implicitly assumes that either we have a small alphabet quite typically 256 symbols or that we have an arbitrary amount of memory at our disposal for the creation of the tree. In real life applications one may need to
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3833 encode symbols coming from a much larger alphabet, for e.g. coding integers. We now have to deal not with hundreds of symbols but possibly with millions of symbols. While other algorithms use a space proportional to the number of observed symbols, we here propose one that uses space proportional to the number of frequency classes, which is, quite interestingly, always smaller or equal to the size of the alphabet.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3834 topics={Compression},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3835 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3836
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3837 @INPROCEEDINGS{Pigeon-dcc99,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3838 author = {Pigeon, Steven and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3839 editor = {Society, {IEEE} Computer},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3840 title = {Binary Pseudowavelets and Applications to Bilevel Image Processing},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3841 booktitle = {Data Compression Conference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3842 year = {1999},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3843 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/dcc99.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3844 abstract = {This paper shows the existance of binary pseudowavelets, bases on the binary domain that exhibit some of the properties of wavelets, such as multiresolution reconstruction and compact support. The binary pseudowavelets are defined on _n (binary vectors of length n) and are operated upon with the binary operators logical and and exclusive or. The forward transform, or analysis, is the decomposition of a binary vector into its constituant binary pseudowavelets. Binary pseudowavelets allow multiresolution, progressive reconstruction of binary vectors by using progressively more coefficients in the inverse transform. Binary pseudowavelets bases, being sparse matrices, also provide for fast transforms; moreover pseudowavelets rely on hardware-friendly operations for efficient software and hardware implementation.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3845 topics={Compression},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3846 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3847
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3848 @TECHREPORT{Pigeon-Huffman-TR98,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3849 author = {Pigeon, Steven and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3850 title = {A Memory-Efficient Adaptive Huffman Coding for Very Large Sets of Symbols revisited},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3851 number = {1095},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3852 year = {1998},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3853 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3854 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/TechRep_AdaptativeHuffman2.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3855 abstract = {While algorithm M (presented in A Memory-Efficient Huffman Adaptive Coding Algorithm for Very Large Sets of Symbols, by Steven Pigeon & Yoshua Bengio, Universit{\'{e}} de Montr{\'{e}}al technical report #1081 [1]) converges to the entropy of the signal, it also assumes that the characteristics of the signal are stationary, that is, that they do not change over time and that successive adjustments, ever decreasing in their magnitude, will lead to a reasonable approximation of the entropy. While this is true for some data, it is clearly not true for some other. We present here a modification of the M algorithm that allows negative updates. Negative updates are used to maintain a window over the source. Symbols enter the window at its right and will leave it at its left, after w steps (the window width). The algorithm presented here allows us to update correctly the weights of the symbols in the symbol tree. Here, we will also have negative migration or demotion, while we only had positive migration or promotion in M. This algorithm will be called M+.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3856 topics={Compression},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3857 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3858
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3859 @PHDTHESIS{Pigeon-Phd-2001,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3860 author = {Pigeon, Steven},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3861 keywords = {algorithmes, codes adaptatifs, codes de Golomb, codes universels, Compression de donn{\'{e}}es, compression LZ78, LZW, ondelettes, pseudo-ondelettes},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3862 title = {Contributions {\`{a}} la compression de donn{\'{e}}es},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3863 year = {2001},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3864 school = {Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3865 abstract = {L'objectif de cette th{\`{e}}se est de pr{\'{e}}senter nos contributions {\`{a}} la compression de donn{\'{e}}es. Le texte entier n'est pas consacr{\'{e}} {\`{a}} nos seules contributions. Une large part est consacr{\'{e}}e au mat{\'{e}}riel introductif et {\`{a}} la recension de litt{\'{e}}rature sur les sujets qui sont pertinents {\`{a}} nos contributions. Le premier chapitre de contribution, le chapitre "Contribution au codage des entiers" se concentre sur le probl{\`{e}}me de la g{\'{e}}n{\'{e}}ration de codes efficaces pour les entiers. Le chapitre "Codage Huffman Adaptatif" pr{\'{e}}sente deux nouveaux algorithmes pour la g{\'{e}}n{\'{e}}ration dynamique de codes structur{\'{e}}s en arbre, c'est-{\`{a}}-dire des codes de type Huffman. Le chapitre "LZW avec une perte" explore le probl{\`{e}}me de la compression d'images comportant un petit nombre de couleurs distinctes et propose une extension avec perte d'un algorithme originalement sans perte, LZW. Enfin, le dernier chapitre de contribution, le chapitre "Les pseudo-ondelettes binaires" pr{\'{e}}sente une solution original au probl{\`{e}}me de l'analyse multir{\'{e}}solution des images monochromes, c'est-{\`{a}}-dire des images n'ayant que deux couleurs, conventionnellement noir et blanc. Ce type d'image correspond par exemple aux images textuelles telle que produites par un processus de transmission de type facsimil{\'{e}}.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3866 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3867
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3868 @ARTICLE{Pigeon98,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3869 author = {Pigeon, Steven and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3870 title = {Memory-Efficient Adaptive Huffman Coding},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3871 journal = {Dr. Dobb's Journal},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3872 volume = {290},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3873 year = {1998},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3874 pages = {131--135},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3875 topics={Compression},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3876 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3877
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3878 @INPROCEEDINGS{probnn:2000:ijcnn,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3879 author = {Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3880 title = {Probabilistic Neural Network Models for Sequential Data},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3881 booktitle = {International Joint Conference on Neural Networks 2000},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3882 volume = {V},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3883 year = {2000},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3884 pages = {79--84},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3885 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/81_01.PDF},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3886 abstract = {It has already been shown how Artificial Neural Networks ({ANN}s) can be incorporated into probabilistic models.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3887 In this paper we review some of the approaches which have been proposed to incorporate them into probabilistic
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3888 models of sequential data, such as Hidden {Markov} Models ({HMM}s). We also discuss new developments and new
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3889 ideas in this area, in particular how {ANN}s can be used to model high-dimensional discrete and continuous data to
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3890 deal with the curse of dimensionality, and how the ideas proposed in these models could be applied to statistical
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3891 language modeling to represent longer-term context than allowed by trigram models, while keeping word-order
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3892 information.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3893 topics={Markov},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3894 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3895
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3896 @UNPUBLISHED{pugin+burgoyne+eck+fujinaga:nipsworkshop2007,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3897 author = {Pugin, L. and Burgoyne, J. A. and Eck, Douglas and Fujinaga, I.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3898 title = {Book-adaptive and book-dependant models to accelerate digitalization of early music},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3899 year = {2007},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3900 note = {NIPS 2007 Workshop on Music, Brain and Cognition},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3901 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3902 sourcetype={Workshop},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3903 optkey={""},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3904 optmonth={""},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3905 optannote={""},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3906 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3907
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3908 @INPROCEEDINGS{Rahim-97,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3909 author = {Rahim, Mazin and Bengio, Yoshua and {LeCun}, Yann},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3910 title = {Discriminative feature and model design for automatic speech recognition},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3911 booktitle = {Proceedings of Eurospeech 1997},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3912 year = {1997},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3913 pages = {75--78},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3914 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/rahim-bengio-lecun-97.ps.gz},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3915 abstract = {A system for discriminative feature and model design is presented for automatic speech recognition. Training based on minimum classification error with a single objective function is applied for designing a set of parallel networks performing feature transformation and a set of hidden {Markov} models performing speech recognition. This paper compares the use of linear and non-linear functional transformations when applied to conventional recognition features, such as spectrum or cepstrum. It also provides a framework for integrated feature and model training when using class-specific transformations. Experimental results on telephone-based connected digit recognition are presented.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3916 topics={Speech},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3917 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3918
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3919 @ARTICLE{Rivest-2009,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3920 author = {Rivest, Fran{\c c}ois and Kalaska, John and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3921 title = {Alternative Time Representations in Dopamine Models},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3922 journal = {Journal of Computational Neuroscience},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3923 volume = {28},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3924 number = {1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3925 year = {2009},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3926 pages = {107--130},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3927 abstract = {Dopaminergic neuron activity has been modeled during learning and appetitive behavior, most commonly using the temporal-difference (TD) algorithm. However, a proper representation of elapsed time and of the exact task is usually required for the model to work. Most models use timing elements such as delay-line representations of time that are not biologically realistic for intervals in the range of seconds. The interval-timing literature provides several alternatives. One of them is that timing could emerge from general network dynamics, instead of coming from a dedicated circuit. Here, we present a general rate-based learning model based on long short-term memory ({LSTM}) networks that learns a time representation when needed. Using a na{\"{\i}}ve network learning its environment in conjunction with TD, we reproduce dopamine activity in appetitive trace conditioning with a constant CS-US interval, including probe trials with unexpected delays. The proposed model learns a representation of the environment dynamics in an adaptive biologically plausible framework, without recourse to delay lines or other special-purpose circuits. Instead, the model predicts that the task-dependent representation of time is learned by experience, is encoded in ramp-like changes in single-neuron activity distributed across small neural networks, and reflects a temporal integration mechanism resulting from the inherent dynamics of recurrent loops within the network. The model also reproduces the known finding that trace conditioning is more difficult than delay conditioning and that the learned representation of the task can be highly dependent on the types of trials experienced during training. Finally, it suggests that the phasic dopaminergic signal could facilitate learning in the cortex.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3928 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3929
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3930 @ARTICLE{schmidhuber+gers+eck:2002,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3931 author = {Schmidhuber, Juergen and Gers, F. A. and Eck, Douglas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3932 title = {Learning Nonregular Languages: A Comparison of Simple Recurrent Networks and {LSTM}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3933 journal = {Neural Computation},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3934 volume = {14},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3935 number = {9},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3936 year = {2002},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3937 pages = {2039--2041},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3938 abstract = {In response to Rodriguez' recent article (Rodriguez 2001) we compare the performance of simple recurrent nets and {\em ``Long Short-Term Memory''} ({LSTM}) recurrent nets on context-free and context-sensitive languages.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3939 source={OwnPublication},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3940 sourcetype={Journal},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3941 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3942
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3943 @TECHREPORT{Schwenk-Bengio-97-TR,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3944 author = {Schwenk, Holger and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3945 title = {Adaptive Boosting of Neural Networks for Character Recognition},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3946 number = {\#1072},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3947 year = {1997},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3948 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3949 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/AdaBoostTR.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3950 abstract = {Boosting is a general method for improving the performance of any learning algorithm that consistently generates classifiers which need to perform only slightly better than random guessing. A recently proposed and very promising boosting algorithm is AdaBoost [5]. It has been applied with great success to several benchmark machine learning problems using rather simple learning algorithms [4], in particular decision trees [1, 2, 6]. In this paper we use AdaBoost to improve the performances of neural networks applied to character recognition tasks. We compare training methods based on sampling the training set and weighting the cost function. Our system achieves about 1.4\% error on a data base of online handwritten digits from more than 200 writers. Adaptive boosting of a multi-layer network achieved 2\% error on the UCI Letters offline characters data set.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3951 topics={Boosting,Speech},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3952 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3953
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3954 @INPROCEEDINGS{Schwenk-nips10,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3955 author = {Schwenk, Holger and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3956 title = {Training Methods for Adaptive Boosting of Neural Networks for Character Recognition},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3957 year = {1998},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3958 crossref = {NIPS10-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3959 abstract = {Boosting is a general method for improving the performance of any learning algorithm that consistently generates classifiers which need to perform only slightly better than random guessing. A recently proposed and very promising boosting algorithm is AdaBoost [5]. It has been applied with great success to several benchmark machine learning problems using rather simple learning algorithms [4], in particular decision trees [1, 2, 6]. In this paper we use AdaBoost to improve the performances of neural networks applied to character recognition tasks. We compare training methods based on sampling the training set and weighting the cost function. Our system achieves about 1.4\% error on a data base of online handwritten digits from more than 200 writers. Adaptive boosting of a multi-layer network achieved 2\% error on the UCI Letters offline characters data set.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3960 topics={Boosting,Speech},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3961 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3962
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3963 @ARTICLE{Schwenk2000,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3964 author = {Schwenk, Holger and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3965 title = {Boosting Neural Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3966 journal = {Neural Computation},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3967 volume = {12},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3968 number = {8},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3969 year = {2000},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3970 pages = {1869--1887},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3971 abstract = {Boosting is a general method for improving the performance of learning algorithms. A recently proposed boosting algorithm is AdaBoost. It has been applied with great success to several benchmark machine learning problems using mainly decision trees as base classifiers. In this paper we investigate whether AdaBoost also works as well with neural networks, and we discuss the advantages and drawbacks of di_erent versions of the AdaBoost algorithm. In particular, we compare training methods based on sampling the training set and weighting the cost function. The results suggest that random resampling of the training data is not the main explanation of the success of the improvements brought by AdaBoost. This is in contrast to Bagging which directly aims at reducing variance and for which random resampling is essential to obtain the reduction in generalization error. Our system achieves about 1.4\% error on a data set of online handwritten digits from more than 200 writers. A boosted multi-layer network achieved 1.5\% error on the UCI Letters and 8.1\% error on the UCI satellite data set, which is significantly better than boosted decision trees.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3972 topics={Boosting},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3973 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3974
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3975 @INPROCEEDINGS{secondorder:2001:nips,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3976 author = {Dugas, Charles and Bengio, Yoshua and Belisle, Francois and Nadeau, Claude and Garcia, Rene},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3977 title = {Incorporating Second-Order Functional Knowledge for Better Option Pricing},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3978 year = {2001},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3979 crossref = {NIPS13-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3980 abstract = {Incorporating prior knowledge of a particular task into the architecture of a learning algorithm can greatly improve generalization performance. We study here a case where we know that the function to be learned is non-decreasing in two of its arguments and convex in one of them. For this purpose we propose a class of functions similar to multi-layer neural networks but (1) that has those properties, (2) is a universal approximator of continuous functions with these and other properties. We apply this new class of functions to the task of modeling the price of call options. Experiments show improvements on regressing the price of call options using the new types of function classes that incorporate the a priori constraints.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3981 topics={Finance},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3982 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3983
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3984 @ARTICLE{Sonnenburg+al-2007,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3985 author = {Sonnenburg, Soeren and et al. and Vincent, Pascal},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3986 title = {The Need for Open Source Software in Machine Learning.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3987 year = {2007},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3988 note = {institution: Fraunhofer Publica [http://publica.fraunhofer.de/oai.har] (Germany)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3989 journal = {Journal of Machine Learning Research},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3990 abstract = {all authors: Sonnenburg, S. and Braun, M.L. and Ong, C.S. and Bengio, S. and Bottou, L. and Holmes, G. and {LeCun}, Y. and M{\~{A}}¼ller, K.-R. and Pereira, F. and Rasmussen, C.E. and R{\~{A}}¤tsch, G. and Sch{\~{A}}{\P}lkopf, B. and Smola, A. and Vincent, P. and Weston, J. and Williamson, R.C.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3991
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3992 Open source tools have recently reached a level of maturity which makes them suitable for building large-scale real-world systems. At the same time, the field of machine learning has developed a large body of powerful learning algorithms for diverse applications. However, the true potential of these methods is not used, since existing implementations are not openly shared, resulting in software with low usability, and weak interoperability. We argue that this situation can be significantly improved by increasing incentives for researchers to publish their software under an open source model. Additionally, we outline the problems authors are faced with when trying to publish algorithmic implementations of machine learning methods. We believe that a resource of peer reviewed software accompanied by short articles would be highly valuable to both the machine learning and the general scientific community.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3993 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3994
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3995 @ARTICLE{Takeuchi-Bengio-Kanamori-2002,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3996 author = {Takeuchi, Ichiro and Bengio, Yoshua and Kanamori, Takafumi},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3997 title = {Robust Regression with Asymmetric Heavy-Tail Noise Distributions},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3998 journal = {Neural Computation},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	3999 volume = {14},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4000 number = {10},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4001 year = {2002},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4002 pages = {2469--2496},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4003 abstract = {In the presence of a heavy-tail noise distribution, regression becomes much more difficult. Traditional robust regression methods assume that the noise distribution is symmetric and they down-weight the influence of so-called outliers. When the noise distribution is assymetric these methods yield biased regression estimators. Motivated by data-mining problems for the insurance industry, we propose in this paper a new approach to robust regession that is tailored to deal with the case where the noise distribution is asymmetric. The main idea is to learn most of the parameters of the model using conditional quantile estimators (which are biased but robust etimators of the regression), and to lern a few remaining parameters to combbine and correct these stimators, to unbiasedly minimize the average squared error. Theoritical analysis and experiments show the clear advantages of the approach. Results are on artificial data as well as real insurance data, using both linear and neural-network predictors.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4004 topics={Mining},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4005 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4006
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4007 @ARTICLE{Thierry+al-2008,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4008 author = {Bertin-Mahieux, Thierry and Eck, Douglas and Maillet, Fran{\c c}ois and Lamere, Paul},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4009 title = {Autotagger: A Model For Predicting Social Tags from Acoustic Features on Large Music Databases},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4010 journal = {Journal of New Music Research},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4011 year = {2008},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4012 abstract = {Social tags are user-generated keywords associated with some resource on the Web. In the case of music, social tags have become an important component of "Web 2.0" recommender systems, allowing users to generate playlists based on use-dependent terms such as chill or jogging that have been applied to particular songs. In this paper, we propose a method for predicting these social tags directly from MP3 files. Using a set of 360 classifiers trained using the online ensemble learning algorithm FilterBoost, we map audio features onto social tags collected from the Web. The resulting automatic tags (or autotags) furnish information about music that is otherwise untagged or poorly tagged, allowing for insertion of previously unheard music into a social recommender. This avoids the cold-start problem common in such systems. Autotags can also be used to smooth the tag space from which similarities and
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4013 recommendations are made by providing a set of comparable baseline tags for all tracks in a recommender system. Because the words we learn are the same as those used by people who label their music collections, it is easy to integrate our predictions into existing similarity and prediction methods based on web data.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4014 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4015
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4016 @ARTICLE{Thivierge+al-2007,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4017 author = {Thivierge, J. -P. and Rivest, Fran{\c c}ois and Monchi, O},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4018 title = {Spiking Neurons, Dopamine, and Plasticity: Timing Is Everything, But Concentration Also Matters},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4019 journal = {Synapse},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4020 volume = {61},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4021 year = {2007},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4022 pages = {375-390},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4023 abstract = {While both dopamine (DA) ﬂuctuations and spike-timing-dependent plasticity (STDP) are known to inﬂuence long-term corticostriatal plasticity, little attention has been devoted to the interaction between these two fundamental mechanisms. Here, a theoretical framework is proposed to account for experimental results specifying the role of presynaptic activation, postsynaptic activation, and concentrations of extracellular DA in synaptic plasticity. Our starting point was an explicitly-implemented multiplicative rule linking STDP to Michaelis-Menton equations that models the dynamics of extracellular DA ﬂuctuations. This rule captures a wide range of results on conditions leading to long-term potentiation and depression in simulations that manipulate the frequency of induced corticostriatal stimulation and DA release. A well-documented biphasic function relating DA concentrations to synaptic plasticity emerges naturally from simulations involving a multiplicative rule linking DA and neural activity. This biphasic function is found consistently across different neural coding schemes employed (voltage-based vs. spike-based models). By comparison, an additive rule fails to capture these results. The proposed framework is the ﬁrst to generate testable predictions on the dual inﬂuence of DA concentrations and STDP on long-term plasticity, suggesting a way in which the biphasic inﬂuence of DA concentrations can modulate the direction and magnitude of change induced by STDP, and raising the possibility that DA concentrations may inverse the LTP/LTD components of the STDP rule.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4024 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4025
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4026 @TECHREPORT{tonga-tr,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4027 author = {Le Roux, Nicolas and Manzagol, Pierre-Antoine and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4028 title = {Topmoumoute online natural gradient algorithm},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4029 number = {1299},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4030 year = {2007},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4031 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4032 abstract = {Guided by the goal of obtaining an optimization algorithm that is
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4033 both fast and yielding good generalization, we study the descent direction maximizing the decrease in generalization error or the probability of not increasing generalization error. The surprising result is that from both the Bayesian and frequentist perspectives this can yield the natural gradient direction. Although that direction can be very expensive to compute we develop an efficient, general, online approximation to the natural gradient descent which is suited to large scale problems. We report experimental results showing much faster convergence in computation time and in number of iterations with TONGA (Topmoumoute Online natural Gradient Algorithm) than with stochastic gradient descent, even on very large datasets.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4034 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4035
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4036 @TECHREPORT{TR1197,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4037 author = {Vincent, Pascal and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4038 title = {K-Local Hyperplane and Convex Distance Nearest Neighbor Algorithms},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4039 number = {1197},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4040 year = {2001},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4041 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4042 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/TR1197.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4043 abstract = {Guided by an initial idea of building a complex (non linear) decision surface with maximal local margin in input space, we give a possible geometrical intuition as to why K-Nearest Neighbor ({KNN}) algorithms often perform more poorly than {SVM}s on classification tasks. We then propose modified K-Nearest Neighbor algorithms to overcome the perceived problem. The approach is similar in spirit to Tangent Distance, but with invariances inferred from the local neighborhood rather than prior knowledge. Experimental results on real world classification tasks suggest that the modified {KNN} algorithms often give a dramatic improvement over standard {KNN} and perform as well or better than {SVM}s.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4044 topics={Kernel},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4045 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4046
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4047 @TECHREPORT{TR1198,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4048 author = {Takeuchi, Ichiro and Bengio, Yoshua and Kanamori, Takafumi},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4049 title = {Robust Regression with Asymmetric Heavy-Tail Noise},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4050 number = {1198},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4051 year = {2001},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4052 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4053 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/TR1198.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4054 abstract = {In the presence of a heavy-tail noise distribution, regression becomes much more difficult. Traditional robust regression methods assume that the noise distribution is symmetric and they downweight the influence of so-called outliers. When the noise distribution is asymmetric these methods yield strongly biased regression estimators. Motivated by data-mining problems for the insurance industry, we propose in this paper a new approach to robust regression that is tailored to deal with the case where the noise distribution is asymmetric. The main idea is to learn most of the parameters of the model using conditional quantile estimators (which are biased but robust estimators of the regression), and to learn a few remaining parameters to combine and correct these estimators, to minimize the average squared error. Theoretical analysis and experiments show the clear advantages of the approach. Results are on artificial data as well as real insurance data, using both linear and neural-network predictors.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4055 topics={Mining},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4056 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4057
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4058 @TECHREPORT{TR1199,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4059 author = {Chapados, Nicolas and Bengio, Yoshua and Vincent, Pascal and Ghosn, Joumana and Dugas, Charles and Takeuchi, Ichiro and Meng, Linyan},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4060 title = {Estimating Car Insurance Premia: a Case Study in High-Dimensional Data Inference},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4061 number = {1199},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4062 year = {2001},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4063 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4064 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/TR1199.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4065 abstract = {Estimating insurance premia from data is a difficult regression problem for several reasons: the large number of variables, many of which are discrete, and the very peculiar shape of the noise distribution, asymmetric with fat tails, with a large majority zeros and a few unreliable and very large values. We introduce a methodology for estimating insurance premia that has been applied in the car insurance industry. It is based on mixtures of specialized neural networks, in order to reduce the effect of outliers on the estimation. Statistical comparisons with several different alternatives, including decision trees and generalized linear models show that the proposed method is significantly more precise, allowing to identify the least and most risky contracts, and reducing the median premium by charging more to the most risky customers.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4066 topics={HighDimensional,Mining},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4067 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4068
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4069 @TECHREPORT{TR1200,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4070 author = {Bengio, Yoshua and Chapados, Nicolas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4071 title = {Extending Metric-Based Model Selection and Regularization in the Absence of Unlabeled Data},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4072 number = {1200},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4073 year = {2001},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4074 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4075 url = {http://www.iro.umontreal.ca/lisa/pointeurs/TR1200.ps},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4076 abstract = {Metric-based methods have recently been introduced for model selection and regularization, often yielding very significant improvements over all the alternatives tried (including cross-validation). However, these methods require a large set of unlabeled data, which is not always available in many applications. In this paper we extend these methods (TRI, ADJ and ADA) to the case where no unlabeled data is available. The extended methods (xTRI, xADJ, xADA) use a model of the input density directly estimated from the training set. The intuition is that the main reason why the above methods work well is that they make sure that the learned function behaves similarly on the training points and on neighboring points. The experiments are based on estimating a simple non-parametric density model. They show that the extended methods perform comparably to the originals even though no unlabeled data is used.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4077 topics={ModelSelection,Finance},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4078 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4079
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4080 @TECHREPORT{TR1215,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4081 author = {Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4082 title = {New Distributed Probabilistic Language Models},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4083 number = {1215},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4084 year = {2002},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4085 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4086 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/TR1215.ps},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4087 abstract = {Our previous work on statistical modeling introduced the use of probabilistic feedforward neural networks with shared parameters in order to help dealing with the curse of dimensionality. This work started with the motivation to speed up the above model and to take advantage of prior knowledge e.g., in WordNet or in syntactically labeled data sets, and to better deal with polysemy. With the objective of reaching these goals, we present here a series of new statistical language models, most of which are yet untested.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4088 topics={Markov,Language,Unsupervised},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4089 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4090
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4091 @TECHREPORT{TR1216,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4092 author = {Bengio, Yoshua and S{\'{e}}n{\'{e}}cal, Jean-S{\'{e}}bastien},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4093 title = {Quick Training of Probabilistic Neural Nets by Importance Sampling},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4094 number = {1216},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4095 year = {2002},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4096 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4097 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/tr1216.ps},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4098 abstract = {Our previous work on statistical language modeling introduced the use of probabilistic feedforward neural networks to help dealing with the curse of dimensionality. Training this model by maximum likelihood however requires for each example to perform as many network passes as there are words in the vocabulary. Inspired by the contrastive divergence model, we proposed and evaluate sampling-based methods which require network passes only for the observed positive example and a few sampled negative example words. A very significant speed-up is obtained with an adaptive importance sampling.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4099 topics={Markov,Language,Unsupervised},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4100 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4101
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4102 @TECHREPORT{TR1231,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4103 author = {Bengio, Yoshua and Kermorvant, Christopher},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4104 title = {Extracting Hidden Sense Probabilities from Bitexts},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4105 number = {1231},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4106 year = {2003},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4107 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4108 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/TR1231.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4109 abstract = {We propose a probabilistic model that is inspired by Diab & Resniks algorithm to extract disambiguation information from aligned bilingual texts. Like Diab & Resniks, the proposed model uses WordNet and the fact that word ambiguities are not always the same in the two languages. The generative model introduces a dependency between two translated words through a common ancestor inWordNets ontology. Unlike Diab & Resniks algorithm it does not suppose that the translation in the source language has a single meaning.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4110 topics={Language},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4111 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4112
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4113 @TECHREPORT{TR1232,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4114 author = {Bengio, Yoshua and Vincent, Pascal and Paiement, Jean-Fran{\c c}ois},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4115 title = {Learning Eigenfunctions of Similarity: Linking Spectral Clustering and Kernel {PCA}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4116 number = {1232},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4117 year = {2003},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4118 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4119 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/TR1232.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4120 abstract = {In this paper, we show a direct equivalence between spectral clustering and kernel {PCA}, and how both are special cases of a more general learning problem, that of learning the principal eigenfunctions of a kernel, when the functions are from a Hilbert space whose inner product is defined with respect to a density model. This suggests a new approach to unsupervised learning in which abstractions (such as manifolds and clusters) that represent the main features of the data density are extracted. Abstractions discovered at one level can be used to build higher-level abstractions. This paper also discusses how these abstractions can be used to recover a quantitative model of the input density, which is at least useful for evaluative and comparative purposes.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4121 topics={HighDimensional,Kernel,Unsupervised},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4122 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4123
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4124 @TECHREPORT{TR1234,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4125 author = {Bengio, Yoshua and Grandvalet, Yves},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4126 title = {No Unbiased Estimator of the Variance of K-Fold Cros-Validation},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4127 number = {1234},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4128 year = {2003},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4129 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4130 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/TR1234.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4131 abstract = {Most machine learning researchers perform quantitative experiments to estimate generalization error and compare the performance of different algorithms (in particular, their proposed algorithm). In order to be able to draw statistically convincing conclusions, it is important for them to also estimate the uncertainty around the error (or error difference) estimate. This paper studies the very commonly used K-fold cross-validation estimator of generalization performance. The main theorem shows that there exists no universal (valid under all distributions) unbiased estimator of the variance of K-fold cross-validation. The analysis that accompanies this result is based on the eigen-decomposition of the covariance matrix of errors, which has only three different eigenvalues corresponding to three degrees of freedom of the matrix and three components of the total variance. This analysis helps to better understand the nature of the problem and how it can make na{\"{\i}}ve estimators (that dont take into account the error correlations due to the overlap between training and test sets) grossly underestimate variance. This is confirmed by numerical experiments in which the three components of the variance are compared when the difficulty of the learning problem and the number of folds are varied.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4132 topics={Comparative},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4133 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4134
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4135 @TECHREPORT{tr1238,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4136 author = {Bengio, Yoshua and Paiement, Jean-Fran{\c c}ois and Vincent, Pascal},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4137 title = {Out-of-Sample Extensions for {LLE}, {I}somap, {MDS}, {E}igenmaps, and Spectral Clustering},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4138 number = {1238},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4139 year = {2003},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4140 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4141 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/tr1238.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4142 abstract = {Several unsupervised learning algorithms based on an eigendecomposition provide either an embedding or a clustering only for given training points, with no straightforward extension for out-of-sample examples short of recomputing eigenvectors. This paper provides algorithms for such an extension for Local Linear Embedding ({LLE}), Isomap, Laplacian Eigenmaps, Multi-Dimensional Scaling (all algorithms which provide lower-dimensional embedding for dimensionality reduction) as well as for Spectral Clustering (which performs non-Gaussian clustering). These extensions stem from a unified framework in which these algorithms are seen as learning eigenfunctions of a kernel. {LLE} and Isomap pose special challenges as the kernel is training-data dependent. Numerical experiments on real data show that the generalizations performed have a level of error comparable to the variability of the embedding algorithms to the choice of training data.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4143 topics={HighDimensional,Kernel,Unsupervised},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4144 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4145
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4146 @TECHREPORT{tr1239,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4147 author = {Bengio, Yoshua and Vincent, Pascal and Paiement, Jean-Fran{\c c}ois and Delalleau, Olivier and Ouimet, Marie and Le Roux, Nicolas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4148 title = {Spectral Clustering and Kernel {PCA} are Learning Eigenfunctions},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4149 number = {1239},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4150 year = {2003},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4151 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4152 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/tr1239.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4153 abstract = {In this paper, we show a direct equivalence between spectral clustering and kernel {PCA}, and how both are special cases of a more general learning problem, that of learning the principal eigenfunctions of a kernel, when the functions are from a function space whose scalar product is defined with respect to a density model. This defines a natural mapping for new data points, for methods that only provided an embedding, such as spectral clustering and Laplacian eigenmaps. The analysis hinges on a notion of generalization for embedding algorithms based on the estimation of underlying eigenfunctions, and suggests ways to improve this generalization by smoothing the data empirical distribution.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4154 topics={HighDimensional,Kernel,Unsupervised},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4155 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4156
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4157 @TECHREPORT{tr1240,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4158 author = {Vincent, Pascal and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4159 title = {Locally Weighted Full Covariance Gaussian Density Estimation},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4160 number = {1240},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4161 year = {2003},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4162 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4163 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/tr1240.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4164 abstract = {We describe an interesting application of the principle of local learning to density estimation. Locally weighted fitting of a Gaussian with a regularized full covariance matrix yields a density estimator which displays improved behavior in the case where much of the probability mass is concentrated along a low dimensional manifold. While the proposed estimator is not guaranteed to integrate to 1 with a finite sample size, we prove asymptotic convergence to the true density. Experimental results illustrating the advantages of this estimator over classic non-parametric estimators are presented.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4165 topics={HighDimensional,Kernel,Unsupervised},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4166 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4167
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4168 @TECHREPORT{tr1247,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4169 author = {Bengio, Yoshua and Delalleau, Olivier and Le Roux, Nicolas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4170 title = {Efficient Non-Parametric Function Induction in Semi-Supervised Learning},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4171 number = {1247},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4172 year = {2004},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4173 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4174 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/tr1247.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4175 abstract = {There has been an increase of interest for semi-supervised learning recently, because of the many datasets with large amounts of unlabeled examples and only a few labeled ones. This paper follows up on proposed non-parametric algorithms which provide an estimated continuous label for the given unlabeled examples. It extends them to function induction algorithms that correspond to the minimization of a regularization criterion applied to an out-of-sample example, and happens to have the form of a Parzen windows regressor. The advantage of the extension is that it allows predicting the label for a new example without having to solve again a linear system of dimension n (the number of unlabeled and labeled training examples), which can cost O(n^3). Experiments show that the extension works well, in the sense of predicting a label close to the one that would have been obtained if the test example had been included in the unlabeled set. This relatively efficient function induction procedure can also be used when n is large to approximate the solution by writing it only in terms of a kernel expansion with m << n terms, and reducing the linear system to m equations in m unknowns.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4176 topics={Kernel,Unsupervised},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4177 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4178
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4179 @TECHREPORT{tr1250,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4180 author = {Bengio, Yoshua and Monperrus, Martin},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4181 title = {Discovering shared structure in manifold learning},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4182 number = {1250},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4183 year = {2004},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4184 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4185 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/tr-tangent.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4186 abstract = {We claim and present arguments to the effect that a large class of manifold learning algorithms that are essentially local will suffer from at least four generic problems associated with (1) noise in the data, (2) curvature of the manifold, (3) dimensionality of the manifold, and (4) the presence of many manifolds with little data per manifold. This analysis suggests non-local manifold learning algorithms which attempt to discover shared structure in the tangent planes at different positions. A criterion for such an algorithm is proposed and experiments estimating a tangent plane prediction function are presented. The function has parameters that are shared across space rather than estimated based on the local neighborhood, as in current non-parametric manifold learning algorithms. The results show clearly the advantages of this approach with respect to local manifold learning algorithms.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4187 topics={HighDimensional,Kernel,Unsupervised},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4188 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4189
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4190 @TECHREPORT{tr1252,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4191 author = {Bengio, Yoshua and Larochelle, Hugo},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4192 title = {Implantation et analyse d'un mod{\`{e}}le graphique {\`{a}} entra{\^{\i}}nement supervis{\'{e}}, semi-supervis{\'{e}} et non-supervis{\'{e}} pour la d{\'{e}}sambigu{\"{\i}}sation s{\'{e}}mantique},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4193 number = {1252},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4194 year = {2004},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4195 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4196 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/tr1252.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4197 abstract = {La d{\'{e}}sambigu{\"{\i}}sation s{\'{e}}mantique est un sujet qui suscite beaucoup dint{\'{e}}r{\^{e}}t dans la communaut{\'{e}} scientifique en apprentissage automatique. Quoique cette t{\^{a}}che ait {\'{e}}t{\'{e}} abord{\'{e}}e depuis les d{\'{e}}buts du traitement automatique de la langue, peu de progr{\`{e}}s ont {\'{e}}t{\'{e}} accomplis jusqu{\`{a}} maintenant. Nous pr{\'{e}}sentons ici une application de d{\'{e}}sambigu{\"{\i}}sation bas{\'{e}}e sur un mod{\`{e}}le graphique probabiliste. Ce mod{\`{e}}le a {\'{e}}t{\'{e}} appris sur des donn{\'{e}}es {\'{e}}tiquet{\'{e}}es, non-{\'{e}}tiquet{\'{e}}es, et sur la hi{\'{e}}rarchie WordNet. Avec peu dexamples dapprentissage, ses performances sont comparables {\`{a}} celles de lalgorithme de Bayes na{\"{\i}}f. Il pourrait {\'{e}}ventuellement {\^{e}}tre adapt{\'{e}} {\`{a}} des corpus bi-textes.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4198 topics={Unsupervised,Language},cat={T},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4199 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4200
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4201 @TECHREPORT{tr1281,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4202 author = {Le Roux, Nicolas and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4203 title = {Continuous Neural Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4204 number = {1281},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4205 year = {2006},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4206 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4207 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/continuous_nnet_tr1281.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4208 abstract = {This article extends neural networks to the case of an uncountable number of hidden units, in several
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4209 ways. In the first approach proposed, a finite parametrization is possible, allowing gradient-based
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4210 learning. While having the same number of parameters as an ordinary neural network, its internal
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4211 structure suggests that it can represent some smooth functions much more compactly. Under mild
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4212 assumptions, we also find better error bounds than with ordinary neural networks. Furthermore, this
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4213 parametrization may help reducing the problem of saturation of the neurons. In a second approach, the
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4214 input-to-hidden weights are fully non-parametric, yielding a kernel machine for which we demonstrate
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4215 a simple kernel formula. Interestingly, the resulting kernel machine can be made hyperparameter-free
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4216 and still generalizes in spite of an absence of explicit regularization.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4217 cat={T},topics={Kernel,HighDimensional},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4218 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4219
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4220 @TECHREPORT{tr1282,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4221 author = {Bengio, Yoshua and Lamblin, Pascal and Popovici, Dan and Larochelle, Hugo},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4222 title = {Greedy Layer-Wise Training of Deep Networks},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4223 number = {1282},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4224 year = {2006},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4225 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4226 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/dbn_supervised_tr1282.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4227 abstract = {Deep multi-layer neural networks have many levels of non-linearities, which allows them to potentially
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4228 represent very compactly highly non-linear and highly-varying functions. However, until recently it
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4229 was not clear how to train such deep networks, since gradient-based optimization starting from random
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4230 initialization appears to often get stuck in poor solutions. Hinton et al. recently introduced a greedy
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4231 layer-wise unsupervised learning algorithm for Deep Belief Networks (DBN), a generative model with
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4232 many layers of hidden causal variables. In the context of the above optimization problem, we study
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4233 this algorithm empirically and explore variants to better understand its success and extend it to cases
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4234 where the inputs are continuous or where the structure of the input distribution is not revealing enough
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4235 about the variable to be predicted in a supervised task.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4236 cat={T},topics={HighDimensional,Unsupervised},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4237 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4238
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4239 @TECHREPORT{tr1283,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4240 author = {Carreau, Julie and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4241 title = {A Hybrid {Pareto} Model for Asymmetric Fat-Tail Data},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4242 number = {1283},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4243 year = {2006},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4244 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4245 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/fat_tails_tr1283.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4246 abstract = {We propose an estimator for the conditional density p(Y \|X) that can adapt for asymmetric heavy tails
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4247 which might depend on X. Such estimators have important applications in finance and insurance. We
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4248 draw from Extreme Value Theory the tools to build a hybrid unimodal density having a parameter
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4249 controlling the heaviness of the upper tail. This hybrid is a Gaussian whose upper tail has been
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4250 replaced by a generalized {Pareto} tail. We use this hybrid in a multi-modal mixture in order to obtain
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4251 a nonparametric density estimator that can easily adapt for heavy tailed data. To obtain a conditional
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4252 density estimator, the parameters of the mixture estimator can be seen as functions of X and these
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4253 functions learned. We show experimentally that this approach better models the conditional density in
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4254 terms of likelihood than compared competing algorithms: conditional mixture models with other types
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4255 of components and multivariate nonparametric models.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4256 cat={T},topics={Unsupervised,Mining},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4257 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4258
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4259 @TECHREPORT{tr1284,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4260 author = {Larochelle, Hugo and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4261 title = {Distributed Representation Prediction for Generalization to New Words},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4262 number = {1284},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4263 year = {2006},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4264 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4265 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/dist_rep_pred_tr1284.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4266 abstract = {Learning distributed representations of symbols (e.g. words) has been used in several Natural Language Processing
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4267 systems. Such representations can capture semantic or syntactic similarities between words, which permit to fight
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4268 the curse of dimensionality when considering sequences of such words. Unfortunately, because these representations
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4269 are learned only for a previously determined vocabulary of words, it is not clear how to obtain representations
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4270 for new words. We present here an approach which gets around this problem by considering the distributed representations
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4271 as predictions from low-level or domain-knowledge features of words. We report experiments on a Part
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4272 Of Speech tagging task, which demonstrates the success of this approach in learning meaningful representations and
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4273 in providing improved accuracy, especially for new words.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4274 cat={T},topics={HighDimensional,Language},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4275 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4276
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4277 @TECHREPORT{tr1285,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4278 author = {Grandvalet, Yves and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4279 title = {Hypothesis Testing for Cross-Validation},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4280 number = {1285},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4281 year = {2006},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4282 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4283 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/xv_rho_stat_tr1285.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4284 abstract = {K-fold cross-validation produces variable estimates, whose variance cannot be estimated unbiasedly. However, in practice, one would like to
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4285 provide a figure related to the variability of this estimate. The first part
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4286 of this paper lists a series of restrictive assumptions (on the distribution of
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4287 cross-validation residuals) that allow to derive unbiased estimates. We exhibit three such estimates, corresponding to differing assumptions. Their
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4288 similar expressions however entail almost identical empirical behaviors.
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4289 Then, we look for a conservative test for detecting significant differences
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4290 in performances between two algorithms. Our proposal is based on the
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4291 derivation of the form of a t-statistic parametrized by the correlation of
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4292 residuals between each validation set. Its calibration is compared to the
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4293 usual t-test. While the latter is overconfident in concluding that differences are indeed significant, our test is bound to be more skeptical, with
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4294 smaller type-I error.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4295 cat={T},topics={ModelSelection,Comparative},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4296 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4297
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4298 @TECHREPORT{tr1286,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4299 author = {Erhan, Dumitru and Bengio, Yoshua and {L'Heureux}, Pierre-Jean and Yue, Shi Yi},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4300 title = {Generalizing to a Zero-Data Task: a Computational Chemistry Case Study},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4301 number = {1286},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4302 year = {2006},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4303 institution = {D{\'{e}}partement d'informatique et recherche op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4304 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/mt_qsar_tr1286.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4305 abstract = {We investigate the problem of learning several tasks simultaneously in order to transfer the acquired
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4306 knowledge to a completely new task for which no training data are available. Assuming that the tasks
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4307 share some representation that we can discover efficiently, such a scenario should lead to a better model of
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4308 the new task, as compared to the model that is learned by only using the knowledge of the new task. We
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4309 have evaluated several supervised learning algorithms in order to discover shared representations among
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4310 the tasks defined in a computational chemistry/drug discovery problem. We have cast the problem from
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4311 a statistical learning point of view and set up the general hypotheses that have to be tested in order
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4312 to validate the multi-task learning approach. We have then evaluated the performance of the learning
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4313 algorithms and showed that it is indeed possible to learn a shared representation of the tasks that allows
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4314 to generalize to a new task for which no training data are available. From a theoretical point of view,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4315 our contribution also comprises a modification to the Support Vector Machine algorithm, which can
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4316 produce state-of-the-art results using multi-task learning concepts at its core. From a practical point
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4317 of view, our contribution is that this algorithm can be readily used by pharmaceutical companies for
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4318 virtual screening campaigns.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4319 cat={T},topics={MultiTask,Kernel,Bioinformatic},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4320 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4321
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4322 @INPROCEEDINGS{Turian+al-2009,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4323 author = {Turian, Joseph and Bergstra, James and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4324 title = {Quadratic Features and Deep Architectures for Chunking},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4325 booktitle = {North American Chapter of the Association for Computational Linguistics - Human Language Technologies (NAACL HLT)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4326 year = {2009},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4327 abstract = {We experiment with several chunking models. Deeper architectures achieve better generalization. Quadratic filters, a simplification of theoretical model of V1 complex cells, reliably increase accuracy. In fact, logistic regression with quadratic filters outperforms a standard single hidden layer neural network. Adding quadratic filters to logistic regression is almost as effective as feature engineering. Despite predicting each output label independently, our model is competitive with ones that use previous decisions.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4328 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4329
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4330 @INPROCEEDINGS{Turian+al-2010,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4331 author = {Turian, Joseph and Ratinov, Lev and Bengio, Yoshua and Roth, Dan},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4332 title = {A preliminary evaluation of word representations for named-entity recognition},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4333 booktitle = {NIPS Workshop on Grammar Induction, Representation of Language and Language Learning},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4334 year = {2009},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4335 url = {http://www.iro.umontreal.ca/~lisa/pointeurs/wordrepresentations-ner.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4336 abstract = {We use different word representations as word features for a named-entity recognition (NER) system with a linear model. This work is part of a larger empirical survey, evaluating different word representations on different NLP tasks. We evaluate Brown clusters, Collobert and Weston (2008) embeddings, and HLBL (Mnih & Hinton, 2009) embeddings of words. All three representations improve accuracy on NER, with the Brown clusters providing a larger improvement than the two embeddings, and the HLBL embeddings more than the Collobert and Weston (2008) embeddings. We also discuss some of the practical issues in using embeddings as features. Brown clusters are simpler than embeddings because they require less hyperparameter tuning.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4337 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4338
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4339 @INPROCEEDINGS{Turian+Ratinov+Bengio-2010,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4340 author = {Turian, Joseph and Ratinov, Lev and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4341 title = {Word representations: A simple and general method for semi-supervised learning},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4342 booktitle = {Association for Computational Linguistics(ACL2010)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4343 year = {2010}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4344 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4345
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4346 @INPROCEEDINGS{Vincent-Bengio-2003,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4347 author = {Vincent, Pascal and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4348 title = {Manifold Parzen Windows},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4349 year = {2003},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4350 pages = {825--832},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4351 crossref = {NIPS15-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4352 abstract = {The similarity between objects is a fundamental element of many learning algorithms. Most non-parametric methods take this similarity to be fixed, but much recent work has shown the advantages of learning it, in particular to exploit the local invariances in the data or to capture the possibly non-linear manifold on which most of the data lies. We propose a new non-parametric kernel density estimation method which captures the local structure of an underlying manifold through the leading eigenvectors of regularized local covariance matrices. Experiments in density estimation show significant improvements with respect to Parzen density estimators. The density estimators can also be used within Bayes classifiers, yielding classification rates similar to {SVM}s and much superior to the Parzen classifier.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4353 topics={HighDimensional,Kernel,Unsupervised},cat={C},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4354 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4355
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4356 @TECHREPORT{Vincent-TR1316,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4357 author = {Vincent, Pascal and Larochelle, Hugo and Bengio, Yoshua and Manzagol, Pierre-Antoine},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4358 title = {Extracting and Composing Robust Features with Denoising Autoencoders},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4359 number = {1316},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4360 year = {2008},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4361 institution = {D{\'{e}}partement d'Informatique et Recherche Op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4362 url = {http://www.iro.umontreal.ca/~vincentp/Publications/denoising_autoencoders_tr1316.pdf},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4363 abstract = {Previous work has shown that the difficulties in learning deep generative or discriminative models can be overcome by an initial unsupervised learning step that maps inputs to useful intermediate representations. We introduce and motivate a new training principle for unsupervised learning of a representation based on the idea of making the learned representations robust to partial corruption of the input pattern. This approach can be used to train autoencoders, and these denoising autoencoders can be stacked to initialize deep architectures. The algorithm can be motivated from a manifold learning and information theoretic perspective or from a generative model perspective. Comparative experiments clearly show the surprising advantage of corrupting the input of autoencoders on a pattern classification benchmark suite.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4364 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4365
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4366 @PHDTHESIS{Vincent2003,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4367 author = {Vincent, Pascal},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4368 title = {Mod{\`{e}}les {\`{a}} Noyaux {\`{a}} Structure Locale},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4369 year = {2003},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4370 school = {D{\'{e}}partement d'Informatique et Recherche Op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4371 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4372
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4373 @ARTICLE{vincent:2001,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4374 author = {Vincent, Pascal and Bengio, Yoshua},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4375 title = {Kernel Matching Pursuit},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4376 journal = {Machine Learning},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4377 year = {2001},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4378 abstract = {We show how Matching Pursuit can be used to build kernel-based solutions to machine-learning problems while keeping control of the sparsity of the solution, and how it can be extended to use non-squared error loss functions. We also deriveMDL motivated generalization bounds for this type of algorithm. Finally, links to boosting algorithms and {RBF} training procedures, as well as extensive experimental comparison with {SVM}s are given, showing comparable results with typically sparser models.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4379 topics={HighDimensional,Kernel},cat={J},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4380 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4381
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4382 @INPROCEEDINGS{VincentPLarochelleH2008,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4383 author = {Vincent, Pascal and Larochelle, Hugo and Bengio, Yoshua and Manzagol, Pierre-Antoine},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4384 title = {Extracting and Composing Robust Features with Denoising Autoencoders},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4385 year = {2008},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4386 pages = {1096--1103},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4387 crossref = {ICML08-shorter},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4388 abstract = {Recently, many applications for Restricted {Boltzmann} Machines (RBMs) have been developed for a large variety of learning problems. However, RBMs are usually used as feature extractors for another learning algorithm or to provide a good initialization
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4389 for deep feed-forward neural network classifiers, and are not considered as a standalone solution to classification problems. In
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4390 this paper, we argue that RBMs provide a self-contained framework for deriving competitive non-linear classifiers. We present an evaluation of different learning algorithms for
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4391 RBMs which aim at introducing a discriminative component to RBM training and improve their performance as classifiers. This
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4392 approach is simple in that RBMs are used directly to build a classifier, rather than as a stepping stone. Finally, we demonstrate how discriminative RBMs can also be successfully employed in a semi-supervised setting.}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4393 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4394
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4395 @TECHREPORT{visualization_techreport,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4396 author = {Erhan, Dumitru and Bengio, Yoshua and Courville, Aaron and Vincent, Pascal},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4397 title = {Visualizing Higher-Layer Features of a Deep Network},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4398 number = {1341},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4399 year = {2009},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4400 institution = {University of Montreal},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4401 abstract = {Deep architectures have demonstrated state-of-the-art results in a variety of
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4402 settings, especially with vision datasets. Beyond the model definitions and the quantitative analyses, there is a need for qualitative comparisons of the solutions learned by various deep architectures. The goal of this paper is to find good qualitative interpretations of high level features represented by such models. To this end, we contrast and compare several techniques applied on Stacked Denoising Autoencoders and Deep Belief Networks, trained on several vision datasets. We show that, perhaps counter-intuitively, such interpretation is possible at the unit level, that it is simple to accomplish and that the results are consistent across various techniques. We hope that such techniques will allow researchers in deep architectures to understand more of how and why deep architectures work}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4403 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4404
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4405 @INPROCEEDINGS{xAISTATS2009-short,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4406 title = {Proc. AISTATS'2009},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4407 booktitle = {Proc. AISTATS'2009},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4408 year = {2009}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4409 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4410
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4411
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4412 @MISC{Yoshua+al-snowbird-2008,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4413 author = {Bengio, Yoshua and Larochelle, Hugo and Turian, Joseph},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4414 title = {Deep Woods},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4415 year = {2008},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4416 howpublished = {Poster presented at the Learning@Snowbird Workshop, Snowbird, USA, 2008}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4417 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4418
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4419 @ARTICLE{Zaccaro-et-al-2005,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4420 author = {Zaccaro, Maria Clara and Boon, Hong and Pattarawarapan, Mookda and Xia, Zebin and Caron, Antoine and {L'Heureux}, Pierre-Jean and Bengio, Yoshua and Burgess, Kevin and Saragori, H. Uri},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4421 title = {Selective Small Molecule Peptidomimetic Ligands of TrkC and TrkA Receptors Afford Discrete or Complete Neurotrophic Activities},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4422 journal = {Chemistry \& Biology},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4423 volume = {12},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4424 number = {9},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4425 year = {2005},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4426 pages = {1015--1028}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4427 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4428
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4429
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4430
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4431 crossreferenced publications:
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4432 @INPROCEEDINGS{ICML09,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4433 editor = {Bottou, {L{\'{e}}on} and Littman, Michael},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4434 title = {Proceedings of the Twenty-sixth International Conference on Machine Learning (ICML'09)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4435 booktitle = {Proceedings of the Twenty-sixth International Conference on Machine Learning (ICML'09)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4436 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4437 publisher = {ACM}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4438 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4439
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4440 @INPROCEEDINGS{NIPS7,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4441 editor = {Tesauro, G. and Touretzky, D. S. and Leen, T. K.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4442 title = {Advances in Neural Information Processing Systems 7 (NIPS'94)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4443 booktitle = {Advances in Neural Information Processing Systems 7 (NIPS'94)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4444 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4445 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4446 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4447
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4448 @INPROCEEDINGS{NIPS6,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4449 editor = {Cowan, J. D. and Tesauro, G. and Alspector, J.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4450 title = {Advances in Neural Information Processing Systems 6 (NIPS'93)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4451 booktitle = {Advances in Neural Information Processing Systems 6 (NIPS'93)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4452 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4453 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4454 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4455
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4456 @INPROCEEDINGS{NIPS8,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4457 editor = {Touretzky, D. S. and Mozer, M. and Hasselmo, M.E.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4458 title = {Advances in Neural Information Processing Systems 8 (NIPS'95)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4459 booktitle = {Advances in Neural Information Processing Systems 8 (NIPS'95)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4460 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4461 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4462 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4463
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4464 @INPROCEEDINGS{NIPS19,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4465 editor = {{Sch{\"{o}}lkopf}, Bernhard and Platt, John and Hoffman, Thomas},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4466 title = {Advances in Neural Information Processing Systems 19 (NIPS'06)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4467 booktitle = {Advances in Neural Information Processing Systems 19 (NIPS'06)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4468 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4469 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4470 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4471
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4472 @INPROCEEDINGS{NIPS10,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4473 editor = {Jordan, M.I. and Kearns, M.J. and Solla, S.A.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4474 title = {Advances in Neural Information Processing Systems 10 (NIPS'97)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4475 booktitle = {Advances in Neural Information Processing Systems 10 (NIPS'97)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4476 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4477 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4478 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4479
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4480 @INPROCEEDINGS{NIPS1,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4481 editor = {Touretzky, D. S.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4482 title = {Advances in Neural Information Processing Systems 1 (NIPS'88)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4483 booktitle = {Advances in Neural Information Processing Systems 1 (NIPS'88)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4484 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4485 publisher = {Morgan Kaufmann}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4486 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4487
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4488 @INPROCEEDINGS{NIPS2,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4489 editor = {Touretzky, D. S.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4490 title = {Advances in Neural Information Processing Systems 2 (NIPS'89)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4491 booktitle = {Advances in Neural Information Processing Systems 2 (NIPS'89)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4492 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4493 publisher = {Morgan Kaufmann}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4494 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4495
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4496 @INPROCEEDINGS{NIPS4,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4497 editor = {Moody, J. E. and Hanson, S. J. and Lipmann, R. P.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4498 title = {Advances in Neural Information Processing Systems 4 (NIPS'91)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4499 booktitle = {Advances in Neural Information Processing Systems 4 (NIPS'91)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4500 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4501 publisher = {Morgan Kaufmann}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4502 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4503
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4504 @INPROCEEDINGS{NIPS12,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4505 editor = {Solla, S.A. and Leen, T. K.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4506 title = {Advances in Neural Information Processing Systems 12 (NIPS'99)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4507 booktitle = {Advances in Neural Information Processing Systems 12 (NIPS'99)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4508 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4509 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4510 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4511
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4512 @INPROCEEDINGS{NIPS16,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4513 editor = {Becker, S. and Saul, L. and {Sch{\"{o}}lkopf}, Bernhard},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4514 title = {Advances in Neural Information Processing Systems 16 (NIPS'03)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4515 booktitle = {Advances in Neural Information Processing Systems 16 (NIPS'03)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4516 year = {-1}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4517 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4518
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4519 @INPROCEEDINGS{NIPS22,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4520 editor = {Bengio, Yoshua and Schuurmans, Dale and Williams, Christopher and Lafferty, John and Culotta, Aron},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4521 title = {Advances in Neural Information Processing Systems 22 (NIPS'09)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4522 booktitle = {Advances in Neural Information Processing Systems 22 (NIPS'09)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4523 year = {-1}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4524 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4525
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4526 @INPROCEEDINGS{NIPS20,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4527 editor = {Platt, John and Koller, D. and Singer, Yoram and Roweis, S.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4528 title = {Advances in Neural Information Processing Systems 20 (NIPS'07)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4529 booktitle = {Advances in Neural Information Processing Systems 20 (NIPS'07)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4530 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4531 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4532 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4533
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4534 @INPROCEEDINGS{xAISTATS2009,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4535 title = {Proceedings of the Twelfth International Conference on Artificial Intelligence and Statistics (AISTATS 2009)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4536 booktitle = {Proceedings of the Twelfth International Conference on Artificial Intelligence and Statistics (AISTATS 2009)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4537 year = {2009},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4538 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4539
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4540 @INPROCEEDINGS{NIPS9,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4541 editor = {Mozer, M. and Jordan, M.I. and Petsche, T.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4542 title = {Advances in Neural Information Processing Systems 9 (NIPS'96)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4543 booktitle = {Advances in Neural Information Processing Systems 9 (NIPS'96)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4544 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4545 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4546 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4547
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4548 @INPROCEEDINGS{NIPS17,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4549 editor = {Saul, Lawrence K. and Weiss, Yair and Bottou, {L{\'{e}}on}},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4550 title = {Advances in Neural Information Processing Systems 17 (NIPS'04)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4551 booktitle = {Advances in Neural Information Processing Systems 17 (NIPS'04)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4552 year = {-1}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4553 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4554
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4555 @INPROCEEDINGS{ICML08,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4556 editor = {Cohen, William W. and McCallum, Andrew and Roweis, Sam T.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4557 title = {Proceedings of the Twenty-fifth International Conference on Machine Learning (ICML'08)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4558 booktitle = {Proceedings of the Twenty-fifth International Conference on Machine Learning (ICML'08)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4559 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4560 publisher = {ACM}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4561 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4562
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4563 @INPROCEEDINGS{ICML07,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4564 editor = {Ghahramani, Zoubin},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4565 title = {Proceedings of the 24th International Conference on Machine Learning (ICML'07)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4566 booktitle = {Proceedings of the 24th International Conference on Machine Learning (ICML'07)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4567 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4568 publisher = {ACM}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4569 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4570
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4571 @TECHREPORT{DIRO,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4572 title = {DIRO},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4573 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4574 institution = {D{\'{e}}partement d'Informatique et de Recherche Op{\'{e}}rationnelle, Universit{\'{e}} de Montr{\'{e}}al},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4575 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4576
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4577 @INPROCEEDINGS{NIPS18,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4578 editor = {Weiss, Yair and {Sch{\"{o}}lkopf}, Bernhard and Platt, John},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4579 title = {Advances in Neural Information Processing Systems 18 (NIPS'05)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4580 booktitle = {Advances in Neural Information Processing Systems 18 (NIPS'05)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4581 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4582 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4583 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4584
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4585 @INPROCEEDINGS{NIPS13,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4586 editor = {Leen, T. K. and Dietterich, T.G.},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4587 title = {Advances in Neural Information Processing Systems 13 (NIPS'00)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4588 booktitle = {Advances in Neural Information Processing Systems 13 (NIPS'00)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4589 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4590 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4591 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4592
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4593 @INPROCEEDINGS{ICML05,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4594 editor = {Raedt, Luc De and Wrobel, Stefan},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4595 title = {Proceedings of the Twenty-second International Conference on Machine Learning (ICML'05)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4596 booktitle = {Proceedings of the Twenty-second International Conference on Machine Learning (ICML'05)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4597 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4598 publisher = {ACM}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4599 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4600
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4601 @INPROCEEDINGS{ICML06,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4602 editor = {Cohen, William W. and Moore, Andrew},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4603 title = {Proceedings of the Twenty-three International Conference on Machine Learning (ICML'06)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4604 booktitle = {Proceedings of the Twenty-three International Conference on Machine Learning (ICML'06)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4605 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4606 publisher = {ACM}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4607 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4608
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4609 @INPROCEEDINGS{NIPS15,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4610 editor = {Becker, S. and Thrun, Sebastian},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4611 title = {Advances in Neural Information Processing Systems 15 (NIPS'02)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4612 booktitle = {Advances in Neural Information Processing Systems 15 (NIPS'02)},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4613 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4614 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4615 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4616
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4617 @INPROCEEDINGS{ICML01-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4618 title = {ICML'01},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4619 booktitle = {ICML'01},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4620 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4621 publisher = {Morgan Kaufmann}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4622 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4623 @INPROCEEDINGS{ICML02-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4624 title = {ICML'02},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4625 booktitle = {ICML'02},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4626 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4627 publisher = {Morgan Kaufmann}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4628 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4629 @INPROCEEDINGS{ICML03-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4630 title = {ICML'03},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4631 booktitle = {ICML'03},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4632 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4633 publisher = {AAAI Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4634 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4635 @INPROCEEDINGS{ICML04-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4636 title = {ICML'04},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4637 booktitle = {ICML'04},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4638 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4639 publisher = {ACM}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4640 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4641 @INPROCEEDINGS{ICML05-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4642 title = {ICML'05},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4643 booktitle = {ICML'05},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4644 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4645 publisher = {ACM}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4646 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4647 @INPROCEEDINGS{ICML06-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4648 title = {ICML'06},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4649 booktitle = {ICML'06},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4650 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4651 publisher = {ACM}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4652 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4653 @INPROCEEDINGS{ICML07-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4654 title = {ICML'07},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4655 booktitle = {ICML'07},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4656 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4657 publisher = {ACM}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4658 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4659 @INPROCEEDINGS{ICML08-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4660 title = {ICML'08},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4661 booktitle = {ICML'08},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4662 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4663 publisher = {ACM}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4664 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4665 @INPROCEEDINGS{ICML09-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4666 title = {ICML'09},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4667 booktitle = {ICML'09},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4668 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4669 publisher = {ACM}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4670 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4671 @INPROCEEDINGS{ICML96-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4672 title = {ICML'96},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4673 booktitle = {ICML'96},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4674 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4675 publisher = {Morgan Kaufmann}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4676 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4677 @INPROCEEDINGS{ICML97-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4678 title = {ICML'97},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4679 booktitle = {ICML'97},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4680 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4681 publisher = {Morgan Kaufmann}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4682 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4683 @INPROCEEDINGS{ICML98-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4684 title = {ICML'98},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4685 booktitle = {ICML'98},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4686 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4687 publisher = {Morgan Kaufmann}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4688 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4689 @INPROCEEDINGS{ICML99-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4690 title = {ICML'99},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4691 booktitle = {ICML'99},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4692 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4693 publisher = {Morgan Kaufmann}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4694 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4695 @INPROCEEDINGS{NIPS1-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4696 title = {NIPS'88},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4697 booktitle = {NIPS 1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4698 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4699 publisher = {Morgan Kaufmann}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4700 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4701 @INPROCEEDINGS{NIPS10-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4702 title = {NIPS'97},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4703 booktitle = {NIPS 10},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4704 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4705 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4706 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4707 @INPROCEEDINGS{NIPS11-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4708 title = {NIPS'98},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4709 booktitle = {NIPS 11},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4710 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4711 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4712 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4713 @INPROCEEDINGS{NIPS12-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4714 title = {NIPS'99},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4715 booktitle = {NIPS 12},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4716 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4717 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4718 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4719 @INPROCEEDINGS{NIPS13-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4720 title = {NIPS'00},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4721 booktitle = {NIPS 13},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4722 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4723 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4724 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4725 @INPROCEEDINGS{NIPS14-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4726 title = {NIPS'01},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4727 booktitle = {NIPS 14},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4728 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4729 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4730 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4731 @INPROCEEDINGS{NIPS15-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4732 title = {NIPS'02},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4733 booktitle = {NIPS 15},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4734 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4735 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4736 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4737 @INPROCEEDINGS{NIPS16-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4738 title = {NIPS'03},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4739 booktitle = {NIPS 16},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4740 year = {-1}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4741 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4742 @INPROCEEDINGS{NIPS17-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4743 title = {NIPS'04},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4744 booktitle = {NIPS 17},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4745 year = {-1}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4746 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4747 @INPROCEEDINGS{NIPS18-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4748 title = {NIPS'05},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4749 booktitle = {NIPS 18},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4750 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4751 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4752 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4753 @INPROCEEDINGS{NIPS19-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4754 title = {NIPS'06},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4755 booktitle = {NIPS 19},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4756 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4757 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4758 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4759 @INPROCEEDINGS{NIPS2-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4760 title = {NIPS'89},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4761 booktitle = {NIPS 2},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4762 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4763 publisher = {Morgan Kaufmann}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4764 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4765 @INPROCEEDINGS{NIPS20-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4766 title = {NIPS'07},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4767 booktitle = {NIPS 20},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4768 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4769 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4770 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4771 @INPROCEEDINGS{NIPS21-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4772 title = {NIPS'08},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4773 booktitle = {NIPS 21},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4774 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4775 publisher = {Nips Foundation (http://books.nips.cc)}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4776 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4777 @INPROCEEDINGS{NIPS22-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4778 title = {NIPS'09},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4779 booktitle = {NIPS 22},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4780 year = {-1}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4781 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4782 @INPROCEEDINGS{NIPS3-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4783 title = {NIPS'90},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4784 booktitle = {NIPS 3},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4785 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4786 publisher = {Morgan Kaufmann}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4787 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4788 @INPROCEEDINGS{NIPS4-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4789 title = {NIPS'91},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4790 booktitle = {NIPS 4},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4791 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4792 publisher = {Morgan Kaufmann}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4793 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4794 @INPROCEEDINGS{NIPS5-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4795 title = {NIPS'92},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4796 booktitle = {NIPS 5},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4797 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4798 publisher = {Morgan Kaufmann}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4799 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4800 @INPROCEEDINGS{NIPS6-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4801 title = {NIPS'93},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4802 booktitle = {NIPS 6},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4803 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4804 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4805 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4806 @INPROCEEDINGS{NIPS7-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4807 title = {NIPS'94},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4808 booktitle = {NIPS 7},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4809 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4810 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4811 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4812 @INPROCEEDINGS{NIPS8-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4813 title = {NIPS'95},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4814 booktitle = {NIPS 8},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4815 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4816 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4817 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4818 @INPROCEEDINGS{NIPS9-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4819 title = {NIPS'96},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4820 booktitle = {NIPS 9},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4821 year = {-1},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4822 publisher = {MIT Press}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4823 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4824 @INPROCEEDINGS{xAISTATS2009-shorter,
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4825 title = {AISTATS'2009},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4826 booktitle = {AISTATS'2009},
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4827 year = {-1}
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4828 }
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4829
b1be957dd1be Added mlj_submission to group every file needed for that. fsavard parents: diff changeset	4830

Mercurial > ift6266

annotate writeup/mlj_submission/aigaion-shorter.bib @ 646:220f5ab44457