Mercurial > ift6266
comparison writeup/nips2010_submission.tex @ 490:d6cf4912abb0
caption + consequent
| author | Yoshua Bengio <bengioy@iro.umontreal.ca> |
|---|---|
| date | Mon, 31 May 2010 22:14:32 -0400 |
| parents | 21787ac4e5a0 |
| children | 19eab4daf212 |
comparison
equal
deleted
inserted
replaced
| 487:21787ac4e5a0 | 490:d6cf4912abb0 |
|---|---|
| 514 comparing the correct digit class with the output class associated with | 514 comparing the correct digit class with the output class associated with |
| 515 the maximum conditional probability among only the digit classes outputs. | 515 the maximum conditional probability among only the digit classes outputs. |
| 516 | 516 |
| 517 \begin{figure}[h] | 517 \begin{figure}[h] |
| 518 \resizebox{.99\textwidth}{!}{\includegraphics{images/error_rates_charts.pdf}}\\ | 518 \resizebox{.99\textwidth}{!}{\includegraphics{images/error_rates_charts.pdf}}\\ |
| 519 \caption{Charts corresponding to table \ref{tab:sda-vs-mlp-vs-humans}. Left: overall results; error bars indicate a 95\% confidence interval. Right: error rates on NIST test digits only, with results from litterature. } | 519 \caption{Left: overall results; error bars indicate a 95\% confidence interval. |
| 520 Right: error rates on NIST test digits only, with results from literature. } | |
| 520 \label{fig:error-rates-charts} | 521 \label{fig:error-rates-charts} |
| 521 \end{figure} | 522 \end{figure} |
| 522 | 523 |
| 523 %\vspace*{-1mm} | 524 %\vspace*{-1mm} |
| 524 %\subsection{Perturbed Training Data More Helpful for SDAE} | 525 %\subsection{Perturbed Training Data More Helpful for SDAE} |
| 554 \fi | 555 \fi |
| 555 | 556 |
| 556 | 557 |
| 557 \begin{figure}[h] | 558 \begin{figure}[h] |
| 558 \resizebox{.99\textwidth}{!}{\includegraphics{images/improvements_charts.pdf}}\\ | 559 \resizebox{.99\textwidth}{!}{\includegraphics{images/improvements_charts.pdf}}\\ |
| 559 \caption{Charts corresponding to tables \ref{tab:perturbation-effect} (left) and \ref{tab:multi-task} (right).} | 560 \caption{Relative improvement in error rate due to self-taught learning. |
| 561 Left: Improvement (or loss, when negative) | |
| 562 induced by out-of-distribution examples (perturbed data). | |
| 563 Right: Improvement (or loss, when negative) induced by multi-task | |
| 564 learning (training on all classes and testing only on either digits, | |
| 565 upper case, or lower-case). The deep learner (SDA) benefits more from | |
| 566 both self-taught learning scenarios, compared to the shallow MLP.} | |
| 560 \label{fig:improvements-charts} | 567 \label{fig:improvements-charts} |
| 561 \end{figure} | 568 \end{figure} |
| 562 | 569 |
| 563 \vspace*{-1mm} | 570 \vspace*{-1mm} |
| 564 \section{Conclusions} | 571 \section{Conclusions} |