Mercurial > pylearn

comparison pylearn/algorithms/linear_regression.py @ 1505:723e2d761985

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
auto white space fix.
author Frederic Bastien <nouiz@nouiz.org>
date Mon, 12 Sep 2011 10:49:15 -0400
parents bf5c0f797161
children
comparison
equal deleted inserted replaced
1504:bf5c0f797161 1505:723e2d761985
24 all_results_dataset=linear_predictor(test_set) # creates a dataset with "output" and "squared_error" field 24 all_results_dataset=linear_predictor(test_set) # creates a dataset with "output" and "squared_error" field
25 outputs = linear_predictor.compute_outputs(inputs) # inputs and outputs are numpy arrays 25 outputs = linear_predictor.compute_outputs(inputs) # inputs and outputs are numpy arrays
26 outputs, errors = linear_predictor.compute_outputs_and_errors(inputs,targets) 26 outputs, errors = linear_predictor.compute_outputs_and_errors(inputs,targets)
27 errors = linear_predictor.compute_errors(inputs,targets) 27 errors = linear_predictor.compute_errors(inputs,targets)
28 mse = linear_predictor.compute_mse(inputs,targets) 28 mse = linear_predictor.compute_mse(inputs,targets)
29 29
30 30
31 31
32 The training_set must have fields "input" and "target". 32 The training_set must have fields "input" and "target".
33 The test_set must have field "input", and needs "target" if 33 The test_set must have field "input", and needs "target" if
34 we want to compute the squared errors. 34 we want to compute the squared errors.
35 35
36 The predictor parameters are obtained analytically from the training set. 36 The predictor parameters are obtained analytically from the training set.
37 37
38 For each (input[t],output[t]) pair in a minibatch,:: 38 For each (input[t],output[t]) pair in a minibatch,::
39 39
40 output_t = b + W * input_t 40 output_t = b + W * input_t
41 41
42 where b and W are obtained by minimizing:: 42 where b and W are obtained by minimizing::
43 43
44 L2_regularizer sum_{ij} W_{ij}^2 + sum_t ||output_t - target_t||^2 44 L2_regularizer sum_{ij} W_{ij}^2 + sum_t ||output_t - target_t||^2
107 107
108 cls.__compiled = True 108 cls.__compiled = True
109 109
110 def __init__(self): 110 def __init__(self):
111 self.compile() 111 self.compile()
112 112
113 class LinearRegressionEquations(LinearPredictorEquations): 113 class LinearRegressionEquations(LinearPredictorEquations):
114 P = LinearPredictorEquations 114 P = LinearPredictorEquations
115 XtX = T.matrix() # (n_inputs+1) x (n_inputs+1) 115 XtX = T.matrix() # (n_inputs+1) x (n_inputs+1)
116 XtY = T.matrix() # (n_inputs+1) x n_outputs 116 XtY = T.matrix() # (n_inputs+1) x n_outputs
117 extended_input = prepend_1_to_each_row(P.inputs) 117 extended_input = prepend_1_to_each_row(P.inputs)
118 new_XtX = T.add(XtX,T.dot(extended_input.T,extended_input)) 118 new_XtX = T.add(XtX,T.dot(extended_input.T,extended_input))
119 new_XtY = T.add(XtY,T.dot(extended_input.T,P.targets)) 119 new_XtY = T.add(XtY,T.dot(extended_input.T,P.targets))
120 120
121 __compiled = False 121 __compiled = False
122 122
123 @classmethod 123 @classmethod
124 def compile(cls, mode="FAST_RUN"): 124 def compile(cls, mode="FAST_RUN"):
125 if cls.__compiled: 125 if cls.__compiled:
126 return 126 return
127 def fn(input_vars,output_vars): 127 def fn(input_vars,output_vars):
154 outputs = self.compute_outputs(inputs) 154 outputs = self.compute_outputs(inputs)
155 return [outputs,self.equations.compute_errors(outputs,targets)] 155 return [outputs,self.equations.compute_errors(outputs,targets)]
156 def compute_mse(self,inputs,targets): 156 def compute_mse(self,inputs,targets):
157 errors = self.compute_errors(inputs,targets) 157 errors = self.compute_errors(inputs,targets)
158 return numpy.sum(errors)/errors.size 158 return numpy.sum(errors)/errors.size
159 159
160 def __call__(self,dataset,output_fieldnames=None,cached_output_dataset=False): 160 def __call__(self,dataset,output_fieldnames=None,cached_output_dataset=False):
161 assert dataset.hasFields(["input"]) 161 assert dataset.hasFields(["input"])
162 if output_fieldnames is None: 162 if output_fieldnames is None:
163 if dataset.hasFields(["target"]): 163 if dataset.hasFields(["target"]):
164 output_fieldnames = ["output","squared_error"] 164 output_fieldnames = ["output","squared_error"]
171 f = self.compute_outputs 171 f = self.compute_outputs
172 elif output_fieldnames == ["output","squared_error"]: 172 elif output_fieldnames == ["output","squared_error"]:
173 f = self.compute_outputs_and_errors 173 f = self.compute_outputs_and_errors
174 else: 174 else:
175 raise ValueError("unknown field(s) in output_fieldnames: "+str(output_fieldnames)) 175 raise ValueError("unknown field(s) in output_fieldnames: "+str(output_fieldnames))
176 176
177 ds=ApplyFunctionDataSet(dataset,f,output_fieldnames) 177 ds=ApplyFunctionDataSet(dataset,f,output_fieldnames)
178 if cached_output_dataset: 178 if cached_output_dataset:
179 return CachedDataSet(ds) 179 return CachedDataSet(ds)
180 else: 180 else:
181 return ds 181 return ds
182 182
183 183
184 def linear_predictor(inputs,params,*otherargs): 184 def linear_predictor(inputs,params,*otherargs):
185 p = LinearPredictor(params) 185 p = LinearPredictor(params)
186 return p.compute_outputs(inputs) 186 return p.compute_outputs(inputs)
187 187
188 #TODO : an online version 188 #TODO : an online version
189 class OnlineLinearRegression():#OnlineLearningAlgorithm): 189 class OnlineLinearRegression():#OnlineLearningAlgorithm):
190 """ 190 """
191 Training can proceed sequentially (with multiple calls to update with 191 Training can proceed sequentially (with multiple calls to update with
193 update the predictor is ready to be used (and optimized for the union 193 update the predictor is ready to be used (and optimized for the union
194 of all the training sets passed to update since construction or since 194 of all the training sets passed to update since construction or since
195 the last call to forget). 195 the last call to forget).
196 """ 196 """
197 pass 197 pass
198
199
200
201