NN is done

source/classifierManager/model/denseNN.py

@@ -29,7 +29,7 @@ import keras as kr
 
 # from keras.utils import plot_model
 from keras.utils.vis_utils import plot_model
-from tensorflow.keras import optimizers
+from keras import optimizers
 
 from classifierManager.model.neuralNetworkInitializer import ModelLSTM
 from classifierManager.model.neuralNetworkInitializer import ModelPerceptron
@@ -58,7 +58,7 @@ class DenseNN:
 
         self.callbacks_list =  [
             kr.callbacks.EarlyStopping(
-                monitor='acc',
+                monitor='val_accuracy',
                 patience=1,
                 ),
             kr.callbacks.ModelCheckpoint(
@@ -129,14 +129,15 @@ class DenseNN:
 
     def compile(self):
         #optimizer = 'rmsprop' #'adam'
-        optimizer = optimizers.RMSprop(lr=0.001, rho=0.9, epsilon=None, decay=0.0)
+        #optimizer = optimizers.RMSprop(lr=0.001, rho=0.9, epsilon=None, decay=0.0)
+        optimizer = optimizers.RMSprop(lr=0.001)
         self.model.compile(optimizer = optimizer, loss = self.loss, metrics = ['accuracy'])
         print("DenseNN - compile() - compile is done")
 
     def fit(self, train_x, train_y, validation_x, validation_y, batchSize=32, epochs=20):
 
         print("DenseNN - fit() - start fitting")
-
+        
         if validation_x is None and validation_y is None:
 
             self.model.fit(x=train_x,
@@ -146,7 +147,7 @@ class DenseNN:
                            verbose=2)
 
         else:
-
+            
             self.model.fit(x=train_x,
                            y=train_y,
                            batch_size=batchSize,

source/classifierManager/script/scriptNeuralNetwork.py

@@ -256,7 +256,7 @@ class ScriptNeuralNetwork(ScriptClassifier):
         
         for _ in range(0, ntime):
             #(accuracy, confusionMatrix, nbEpoch, auc) 
-            result = self.predictionTask(course, whereToCut, cache=cache)
+            result = self.predictionTask(course, whereToCut)
 
             if auc is None:
                 raise ValueError("classifierManager.script.ScriptNeuralNetwork - predictionTaskNTimes : auc is None")
@@ -310,6 +310,8 @@ class ScriptNeuralNetwork(ScriptClassifier):
         accuracies = []
         aucs = []
         epochs = []
+        
+        size = len(self.classifier.nameGroups)
         avgConfusionMatrix = np.zeros((size, size))
         accuraciesPerClass = []
         dictAllCourses = {}
@@ -349,7 +351,7 @@ class ScriptNeuralNetwork(ScriptClassifier):
                 fileResult.write(str(auc))
                 
                 fileResult.write("\n nbEpoch\n")
-                fileResult.write(str(nbEpoch))
+                fileResult.write(str(epoch))
                 
                 
 

source/main/ocr/classificationExperiment.py

@@ -109,7 +109,7 @@ if __name__ == "__main__":
     
     ''' for all periods '''
      
-    classifier.predictionTaskForAllPeriods(ntime=10, cache=False)
+    classifier.predictionTaskForAllPeriods(ntime=10, cache=True)
 
     ''' NN : all courses with hidden layers'''