# HG changeset patch
# User Xavier Glorot <glorotxa@iro.umontreal.ca>
# Date 1244902303 14400
# Node ID ba055d419bcfcc812d0ccfd5c9fa91555f9a828d
# Parent  41761210d16e7d3d374209a3644c8fb038d8090a
change seed initialisation for random object and weigth init for logistic regression

diff -r 41761210d16e -r ba055d419bcf pylearn/algorithms/sandbox/DAA_inputs_groups.py
--- a/pylearn/algorithms/sandbox/DAA_inputs_groups.py	Thu Jun 11 11:37:24 2009 -0400
+++ b/pylearn/algorithms/sandbox/DAA_inputs_groups.py	Sat Jun 13 10:11:43 2009 -0400
@@ -557,7 +557,7 @@
                 self.totalupdate[i] = theano.Method(self.inputs[i],self.totalcost[i],total_grads)
             #
             if self.debugmethod:
-                self.representation[i] = theano.Method(self.inputs[i],self.daaig[i].clean.hidden)
+                self.representation[i] = theano.Method(self.inputs[i],self.daaig[i].clean.hidden_activation)
                 self.reconstruction[i] = theano.Method(self.inputs[i],self.daaig[i].clean.rec)
                 self.validate[i] =theano.Method(self.inputs[i], [self.daaig[i].clean.cost, self.daaig[i].clean.rec])
                 self.noisyinputs[i] =theano.Method(self.inputs[i], noisyout)
@@ -639,8 +639,13 @@
         for i in range(self.depth):
             print '\tLayer = ', i+1
             inst.daaig[i].initialize(reg_coef = reg_coef, noise_level = noise_level,\
-                    noise_level_group = noise_level_group, seed = seed, alloc = alloc)
+                    noise_level_group = noise_level_group, seed = seed + i, alloc = alloc)
         print '\tLayer supervised'
         inst.daaig[-1].initialize()
+        if alloc:
+            inst.daaig[-1].R = numpy.random.RandomState(seed+self.depth)
+            # init the logreg weights
+            inst.daaig[-1].w = inst.daaig[-1].R.uniform(size=inst.daaig[-1].w.shape,\
+                                low = -1/numpy.sqrt(inst.daaig[-2].n_hid), high = 1/numpy.sqrt(inst.daaig[-2].n_hid))
         inst.daaig[-1].l1 = 0
         inst.daaig[-1].l2 = reg_coef #only l2 norm for regularisation to be consitent with the unsup regularisation