Mercurial > ift6266

comparison deep/stacked_dae/v_sylvain/sgd_optimization.py @ 397:5e11dda78995

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Ajout de fonctionnalite pour pouvoir facilement calculer une erreur sur differentes parties de NSIT
author SylvainPL <sylvain.pannetier.lebeuf@umontreal.ca>
date Tue, 27 Apr 2010 20:50:56 -0400
parents 88cb95007670
children 09e1c5872c2b
comparison
equal deleted inserted replaced
396:116b2de2c0a4 397:5e11dda78995
392 if x.dtype=='float64': 392 if x.dtype=='float64':
393 self.classifier.params[idx].value=theano._asarray(copy(x),dtype=theano.config.floatX) 393 self.classifier.params[idx].value=theano._asarray(copy(x),dtype=theano.config.floatX)
394 else: 394 else:
395 self.classifier.params[idx].value=copy(x) 395 self.classifier.params[idx].value=copy(x)
396 396
397 def training_error(self,dataset): 397 def training_error(self,dataset,part=0):
398 # create a function to compute the mistakes that are made by the model 398 # create a function to compute the mistakes that are made by the model
399 # on the validation set, or testing set 399 # on the validation set, or testing set
400 test_model = \ 400 test_model = \
401 theano.function( 401 theano.function(
402 [self.classifier.x,self.classifier.y], self.classifier.errors) 402 [self.classifier.x,self.classifier.y], self.classifier.errors)
403 403 #train
404 iter2 = dataset.train(self.hp.minibatch_size,bufsize=buffersize) 404 if part == 0:
405 iter2 = dataset.train(self.hp.minibatch_size,bufsize=buffersize)
406 name = 'train'
407 #validation
408 if part == 1:
409 iter2 = dataset.valid(self.hp.minibatch_size,bufsize=buffersize)
410 name = 'validation'
411 if part == 2:
412 iter2 = dataset.test(self.hp.minibatch_size,bufsize=buffersize)
413 name = 'test'
405 train_losses2 = [test_model(x,y) for x,y in iter2] 414 train_losses2 = [test_model(x,y) for x,y in iter2]
406 train_score2 = numpy.mean(train_losses2) 415 train_score2 = numpy.mean(train_losses2)
407 print(('The training error is %f')%(train_score2*100.)) 416 print 'On the ' + name + 'dataset'
417 print(('\t the error is %f')%(train_score2*100.))
408 418
409 #To see the prediction of the model, the real answer and the image to judge 419 #To see the prediction of the model, the real answer and the image to judge
410 def see_error(self, dataset): 420 def see_error(self, dataset):
411 import pylab 421 import pylab
412 #The function to know the prediction 422 #The function to know the prediction