[algorithm] fixed precision criterion to stop training

bob/pad/base/algorithm/MLP.py

@@ -38,7 +38,6 @@ class MLP(Algorithm):
     
     # The data
     batch_size = len(training_features[0]) + len(training_features[1])
-    print(batch_size)  
     label_real = numpy.ones((len(training_features[0]), 1), dtype='float64')
     label_attack = numpy.zeros((len(training_features[1]), 1), dtype='float64')
 
@@ -68,12 +67,12 @@ class MLP(Algorithm):
     previous_cost = 0
     current_cost = 1
     precision = 0.001
-    while (n_iter < self.max_iter) or (abs(previous_cost - current_cost) < precision): 
+    while (n_iter < self.max_iter) or (abs(previous_cost - current_cost) > precision): 
       previous_cost = current_cost
       trainer.train(self.mlp, X, Y)
       current_cost = trainer.cost(self.mlp, X, Y)
       n_iter += 1
-      print("Iteration {} -> cost = {} (previous = {})".format(n_iter, trainer.cost(self.mlp, X, Y), previous_cost))
+      #print("Iteration {} -> cost = {} (previous = {})".format(n_iter, trainer.cost(self.mlp, X, Y), previous_cost))
 
     f = bob.io.base.HDF5File(projector_file, 'w')
     self.mlp.save(f)