améliorations rapport et correction bogues dans le code

Code/NeuralNet.py

@@ -13,7 +13,6 @@ import numpy as np
 import time
 import NeuralNode
 import metrics
-import random
 
 # le nom de votre classe
 # NeuralNet pour le modèle Réseaux de Neurones
@@ -21,7 +20,7 @@ import random
 
 class NeuralNet: #nom de la class à changer
 
-    def __init__(self, layers_size, all_labels, **kwargs):
+    def __init__(self, layers_size, all_labels, input_weights=None, **kwargs):
         """
         c'est un Initializer. 
         Vous pouvez passer d'autre paramètres au besoin,
@@ -36,11 +35,11 @@ class NeuralNet: #nom de la class à changer
         # Couches cachées
         for j in range(self.n_hidden_layers):
             for i in range(self.layers_size[j+1]):
-                self.layers[j].append(NeuralNode.NeuralNode(layers_size[j]))
+                self.layers[j].append(NeuralNode.NeuralNode(layers_size[j],input_weights))
         
         # Couche de sortie
         for i in range(layers_size[-1]):
-            self.layers[-1].append(NeuralNode.NeuralNode(layers_size[-2]))
+            self.layers[-1].append(NeuralNode.NeuralNode(layers_size[-2],input_weights))
             
     def __str__(self):
         outstr = ""

Code/NeuralNode.py

@@ -11,10 +11,10 @@ import NeuralNetUtils as nnu
 class NeuralNode:
     def __init__(self, input_size, input_weights=None, **kwargs):
         self.input_size = input_size+1
-        if (input_weights == None):
+        if (input_weights is None):
             self.input_weights = np.random.uniform(-1,1,self.input_size)
         else:
-            self.input_weights = input_weights
+            self.input_weights = np.repeat(input_weights,self.input_size)
             
     def __str__(self):
         return str(self.input_weights)

Code/metrics.py

@@ -41,13 +41,13 @@ def prediction_metrics(cm,obs_labels,pred_labels):
             if (not np.any(np.isnan(myPrecision))):
                 precision.append(myPrecision)
         except:
-            raise
+            myPrecision = 0
         try:
             myRecall = cm[label_num,label_num] / sum(cm[label_num,:])
             if (not np.any(np.isnan(myRecall))):
                 recall.append(myRecall)
         except:
-            raise
+            myRecall = 0
     
     return accuracy, precision, recall