diff deep/stacked_dae/v_sylvain/stacked_dae.py @ 251:02b141a466b4

ajout de fonctionnalite pour different finetune dataset
author SylvainPL <sylvain.pannetier.lebeuf@umontreal.ca>
date Tue, 16 Mar 2010 21:24:30 -0400
parents ecb69e17950b
children c77ffb11f91d
line wrap: on
line diff
--- a/deep/stacked_dae/v_sylvain/stacked_dae.py	Tue Mar 16 21:24:09 2010 -0400
+++ b/deep/stacked_dae/v_sylvain/stacked_dae.py	Tue Mar 16 21:24:30 2010 -0400
@@ -165,9 +165,9 @@
 
 
 class SdA(object):
-    def __init__(self, train_set_x, train_set_y, batch_size, n_ins, 
+    def __init__(self, batch_size, n_ins, 
                  hidden_layers_sizes, n_outs, 
-                 corruption_levels, rng, pretrain_lr, finetune_lr, input_divider=1.0):
+                 corruption_levels, rng, pretrain_lr, finetune_lr):
         # Just to make sure those are not modified somewhere else afterwards
         hidden_layers_sizes = copy.deepcopy(hidden_layers_sizes)
         corruption_levels = copy.deepcopy(corruption_levels)
@@ -190,23 +190,17 @@
         print "n_outs", n_outs
         print "pretrain_lr", pretrain_lr
         print "finetune_lr", finetune_lr
-        print "input_divider", input_divider
         print "----"
 
-        #self.shared_divider = theano.shared(numpy.asarray(input_divider, dtype=theano.config.floatX))
-
         if len(hidden_layers_sizes) < 1 :
             raiseException (' You must have at least one hidden layer ')
 
 
         # allocate symbolic variables for the data
-        ##index   = T.lscalar()    # index to a [mini]batch 
+        #index   = T.lscalar()    # index to a [mini]batch 
         self.x  = T.matrix('x')  # the data is presented as rasterized images
         self.y  = T.ivector('y') # the labels are presented as 1D vector of 
                                  # [int] labels
-        ensemble = T.matrix('ensemble')
-        ensemble_x = T.matrix('ensemble_x')
-        ensemble_y = T.ivector('ensemble_y')
 
         for i in xrange( self.n_layers ):
             # construct the sigmoidal layer
@@ -250,10 +244,15 @@
                 updates[param] = param - gparam * pretrain_lr
             
             # create a function that trains the dA
-            update_fn = theano.function([ensemble], dA_layer.cost, \
-                  updates = updates,
-                  givens = { 
-                     self.x : ensemble})
+            update_fn = theano.function([self.x], dA_layer.cost, \
+                  updates = updates)#,
+            #     givens = { 
+            #         self.x : ensemble})
+            # collect this function into a list
+            #update_fn = theano.function([index], dA_layer.cost, \
+            #      updates = updates,
+            #      givens = { 
+            #         self.x : train_set_x[index*batch_size:(index+1)*batch_size] / self.shared_divider})
             # collect this function into a list
             self.pretrain_functions += [update_fn]
 
@@ -276,18 +275,17 @@
         for param,gparam in zip(self.params, gparams):
             updates[param] = param - gparam*finetune_lr
             
-        self.finetune = theano.function([ensemble_x,ensemble_y], cost, 
-                updates = updates,
-                givens = {
-                  #self.x : train_set_x[index*batch_size:(index+1)*batch_size]/self.shared_divider,
-                  #self.y : train_set_y[index*batch_size:(index+1)*batch_size]} )
-                  self.x : ensemble_x,
-                  self.y : ensemble_y} )
+        self.finetune = theano.function([self.x,self.y], cost, 
+                updates = updates)#,
+        #        givens = {
+        #          self.x : train_set_x[index*batch_size:(index+1)*batch_size]/self.shared_divider,
+        #          self.y : train_set_y[index*batch_size:(index+1)*batch_size]} )
 
         # symbolic variable that points to the number of errors made on the
         # minibatch given by self.x and self.y
 
         self.errors = self.logLayer.errors(self.y)
+        
 
 if __name__ == '__main__':
     import sys