Save

Trying do make the backward to work

include/neural_network/activation_loss/activation_softmax_loss_categorical_crossentropy.hpp

@@ -92,7 +92,7 @@ struct activation_softmax_loss_categorical_crossentropy: loss{
      * @param y_true Vector of true label of data.
      * 
      */
-     bool forward(const panic::tensor::real_matrix& y_pred, const panic::tensor::uint_vector& y_true);
+     bool calculate(const panic::tensor::real_matrix& y_pred, const panic::tensor::uint_vector& y_true) override;
 
     /**
      * @brief forward function to calculate losses
@@ -103,7 +103,7 @@ struct activation_softmax_loss_categorical_crossentropy: loss{
      * @Note Overloaded if one-shot endcoded
      *         is used.
      */
-    bool forward(const panic::tensor::real_matrix& y_pred, const panic::tensor::real_matrix& y_true);
+    bool calculate(const panic::tensor::real_matrix& y_pred, const panic::tensor::real_matrix& y_true) override;
 
 
 

include/neural_network/loss/loss.hpp

@@ -73,6 +73,11 @@ struct loss{
      */
     panic::tensor::real_matrix dinputs;
 
+    /**
+     * @brief Matrix for output of loss function
+     */
+    panic::tensor::real_matrix outputs;
+
 
     /**
      * @brief Default de-constructor

include/neural_network/loss/loss_categorical_crossentropy.hpp

@@ -92,7 +92,7 @@ struct loss_categorical_crossentropy: loss{
      */
     bool backward(
         const panic::tensor::real_matrix& dvalues, 
-        const panic::tensor::uint_vector& y_true);
+        const panic::tensor::uint_vector& y_true) override;
 
     /**
      * @brief backward function to calculate from losses
@@ -105,7 +105,7 @@ struct loss_categorical_crossentropy: loss{
      */
     bool backward(
         const panic::tensor::real_matrix& dvalues, 
-        const panic::tensor::real_matrix& y_true);
+        const panic::tensor::real_matrix& y_true) override;
 
 
 };

src/neural_network/activation_loss/activation_softmax_loss_categorical_crossentropy.cpp

@@ -70,65 +70,6 @@ namespace panic{
 activation_softmax_loss_categorical_crossentropy::activation_softmax_loss_categorical_crossentropy() {
 }
 
-
-//--------------------------------------------------------------------------------------------------------------------------
-// Function Name : panic::neural_network::activation_softmax_loss_categorical_crossentropy.forward
-//
-// Description:
-//		Calculated the forward pass
-//--------------------------------------------------------------------------------------------------------------------------
-bool activation_softmax_loss_categorical_crossentropy::forward(const panic::tensor::real_matrix& y_pred, const panic::tensor::uint_vector& y_true){
-
-	// Output layers activation function
-	if (!activation.forward(y_pred)){
-	 	return false;
-	 } 
-	// Set the output
-	outputs = activation.outputs;
-
-	// calculate the loss value.
-	if (!loss.calculate(outputs, y_true)){
-		return false;
-	}
-	
-	return true;
-}
-
-//--------------------------------------------------------------------------------------------------------------------------
-// Function Name : panic::neural_network::activation_softmax_loss_categorical_crossentropy.forward
-//
-// Description:
-//		Calculated the forward pass
-//--------------------------------------------------------------------------------------------------------------------------
-bool activation_softmax_loss_categorical_crossentropy::forward(const panic::tensor::real_matrix& y_pred, const panic::tensor::real_matrix& y_true){
-
-	// Output layers activation function
-	if (!activation.forward(y_pred)){
-	 	return false;
-	 } 
-	// Set the output
-	outputs = activation.outputs;
-
-	// calculate the loss value.
-	if (!loss.calculate(outputs, y_true)){
-		return false;
-	}
-	
-	return true;
-}
-
-
-
-
-
-
-
-
-
-
-
-
-
 //--------------------------------------------------------------------------------------------------------------------------
 // Function Name : panic::neural_network::activation_softmax_loss_categorical_crossentropy.calculate
 //

src/neural_network/loss/loss.cpp

@@ -77,7 +77,10 @@ bool loss::calculate(
     // Calculate mean loss
     data_loss = panic::math::mean(sample_losses);
 
-    //return calculate_mean();
+    // This is so the model can use every loss functions
+    // It's a problem to get the output when useing an activation+loss function
+    outputs = y_pred;
+
     return true;
 }
 
@@ -97,6 +100,11 @@ bool loss::calculate(
 
     data_loss = panic::math::mean(sample_losses);
 
+
+    // This is so the model can use every loss functions
+    // It's a problem to get the output when useing an activation+loss function
+    outputs = y_pred;
+    
     return true;
 }
 
@@ -127,38 +135,6 @@ bool loss::forward(
     return false;
 }
 
-/*
-
-//--------------------------------------------------------------------------------------------------------------------------
-// Function Name : panic::neural_network::loss::calculate_mean
-//
-// Description:
-//      Calculates the average of all sample loss values.
-//--------------------------------------------------------------------------------------------------------------------------
-bool loss::calculate_mean(){
-
-    if (sample_losses.size() == 0){
-        return false;
-    }
-
-    panic::types::real_t sum =
-        static_cast<panic::types::real_t>(0);
-
-    for (
-        panic::types::uint_t i = 0;
-        i < sample_losses.size();
-        ++i
-    ){
-        sum += sample_losses[i];
-    }
-
-    data_loss =
-        sum /
-        static_cast<panic::types::real_t>(sample_losses.size());
-
-    return true;
-}
-*/
 
 	} // namespace tensor
 } // namespace panic

src/neural_network/model/model.cpp

@@ -450,7 +450,7 @@ bool model::train(const panic::tensor::real_matrix& X_train,
         }
 
         
-        prediction = panic::math::argmax_rowwise(outputs);
+        prediction = panic::math::argmax_rowwise(loss_function->outputs);
 
         comparisons = panic::math::equal(prediction, y_train);
 
@@ -463,6 +463,8 @@ bool model::train(const panic::tensor::real_matrix& X_train,
         }
 
 
+        loss_function.backward(loss_function.outputs, y_train);
+
         //backward();
 
         //optimize();