python - Adding a new term to a built-in loss function in keras

Question

I am working with the below loss function to train the neural network that I am working with:

loss = {'time_prediction': keras.losses.mean_squared_error,
            'decoded_mean': keras.losses.mean_squared_error}

Now, I need to update my loss function by adding a new term to the previous loss function which is not a built-in term in keras. How can I add this new custom function to the previous loss function? I wrote the new term as a function below:

def new_term_loss(X_alpha, X_train):
  return (1/np.size(X_alpha)) * np.sum((X_alpha - X_train)>0)

question from:https://stackoverflow.com/questions/65922117/adding-a-new-term-to-a-built-in-loss-function-in-keras

Answer 1

Waitting for answers