[mysvm] added some debug stuff

bob/pad/base/algorithm/mySVM.py

@@ -13,7 +13,7 @@ class mySVM(Algorithm):
 
   """
 
-  def __init__(self, rescale=True, gamma=0.1, **kwargs):
+  def __init__(self, rescale=True, gamma=0.1, cost=1.0,  **kwargs):
 
     Algorithm.__init__(self,  
                        performs_projection=True,
@@ -22,9 +22,11 @@ class mySVM(Algorithm):
 
     self.rescale = rescale 
     self.gamma = gamma
+    self.cost = cost
     self.machine = None
-    self.trainer = bob.learn.libsvm.Trainer(machine_type='C_SVC', kernel_type='RBF', probability=True)
+    self.trainer = bob.learn.libsvm.Trainer(machine_type='C_SVC', kernel_type='RBF', probability=False)
     setattr(self.trainer, 'gamma', self.gamma)
+    setattr(self.trainer, 'cost', self.cost)
 
 
   def train_projector(self, training_features, projector_file):
@@ -38,24 +40,51 @@ class mySVM(Algorithm):
     # training_features[0] - training features for the REAL class.
     # training_features[1] - training features for the ATTACK class.
     
-    # The data - "positive class only"
+    # The data 
     pos = numpy.array(training_features[0])
     neg = numpy.array(training_features[1])
 
+    for i in range(pos.shape[0]):
+      print("real feature {}:".format(i))
+      print(pos[i])
+
+    print("positive examples shape {}".format(pos.shape))
+    print("negative examples shape {}".format(neg.shape))
+
+    # -----------------------------------------------------
+    # Olegs whole stuff
+    #n_real = pos.shape[0]
+    #n_attack = neg.shape[0]
+    #pos_train = pos[:numpy.int(n_real/2)]
+    #neg_train = neg[:numpy.int(n_real/2)]
+    #pos_cv = pos[numpy.int(n_real/2):2*numpy.int(n_real/2)+1]
+    #neg_cv = neg[numpy.int(n_real/2):2*numpy.int(n_real/2)+1]
+
+    # -----------------------------------------------------
+
+    print("Feature before rescaling")
+    print(pos[0])
+
     if self.rescale:
       for i in range(pos.shape[0]):
         min_value = numpy.min(pos[i])
         max_value = numpy.max(pos[i])
+        #pos[i] = (pos[i] - min_value)/ (max_value - min_value)
         pos[i] = ((2 * (pos[i] - min_value))/ (max_value - min_value)) - 1
       for i in range(neg.shape[0]):
         min_value = numpy.min(neg[i])
         max_value = numpy.max(neg[i])
+        #neg[i] = (neg[i] - min_value)/ (max_value - min_value)
         neg[i] = ((2 * (neg[i] - min_value))/ (max_value - min_value)) - 1
     
+    print("Feature after rescaling")
+    print(pos[0])
+    
     data = [pos, neg]
     
     # train
     self.machine = self.trainer.train(data)
+    print(self.machine.shape)
     f = bob.io.base.HDF5File(projector_file, 'w')
     self.machine.save(f)