Custom loss function in Keras based on the input data

by

I have come across 2 solutions to the question you asked.

  1. You can pass your input tensor as an argument to the custom loss wrapper function.
    def custom_loss(i):

        def loss(y_true, y_pred):
            return K.mean(K.square(y_pred - y_true), axis=-1) + something with i...
        return loss

    def baseline_model():
        # create model
        i = Input(shape=(5,))
        x = Dense(5, kernel_initializer="glorot_uniform", activation='linear')(i)
        o = Dense(1, kernel_initializer="normal", activation='linear')(x)
        model = Model(i, o)
        model.compile(loss=custom_loss(i), optimizer=Adam(lr=0.0005))
        return model

This solution is also mentioned in the accepted answer here

  1. You can pad your label with extra data columns from input and write a custom loss. This is helpful if you just want one/few feature column(s) from your input.
    def custom_loss(data, y_pred):

        y_true = data[:, 0]
        i = data[:, 1]
        return K.mean(K.square(y_pred - y_true), axis=-1) + something with i...


    def baseline_model():
        # create model
        i = Input(shape=(5,))
        x = Dense(5, kernel_initializer="glorot_uniform", activation='linear')(i)
        o = Dense(1, kernel_initializer="normal", activation='linear')(x)
        model = Model(i, o)
        model.compile(loss=custom_loss, optimizer=Adam(lr=0.0005))
        return model


    model.fit(X, np.append(Y_true, X[:, 0], axis =1), batch_size = batch_size, epochs=90, shuffle=True, verbose=1)

This solution can be found also here in this thread.

I have only used the 2nd method when I had to use input feature columns in the loss. I have used the first method with scalar arguments; but I believe a tensor input works as well.

More Related Contents:

  1. When does keras reset an LSTM state?
  2. How to calculate F1 Macro in Keras?
  3. What is the meaning of axis=-1 in keras.argmax?
  4. How can I use a weight matrix created by a trained neural network to make predictions in another file?
  5. Keras, How to get the output of each layer?
  6. How do you create a custom activation function with Keras?
  7. Deep-Learning Nan loss reasons
  8. Failed to get convolution algorithm. This is probably because cuDNN failed to initialize,
  9. Why is TensorFlow 2 much slower than TensorFlow 1?
  10. Many to one and many to many LSTM examples in Keras
  11. NaN loss when training regression network
  12. Can Keras with Tensorflow backend be forced to use CPU or GPU at will?
  13. Can I run Keras model on gpu?
  14. Neural Network LSTM input shape from dataframe
  15. Convert Keras model to C++ [closed]
  16. Keras custom loss function: Accessing current input pattern
  17. How to determine needed memory of Keras model?
  18. Standard implementation of vectorize_sequences
  19. Get class labels from Keras functional model
  20. Loading a trained Keras model and continue training
  21. How to save final model using keras?
  22. How does Keras handle multilabel classification?
  23. How do I create a variable-length input LSTM in Keras?
  24. Unknown initializer: GlorotUniform when loading Keras model
  25. AttributeError: ‘Model’ object has no attribute ‘predict_classes’
  26. How to use fit_generator with multiple inputs
  27. Multiple outputs in Keras
  28. Can Keras deal with input images with different size?
  29. How to implement Grad-CAM on a trained network
  30. How can I convert a trained Tensorflow model to Keras?

Leave a Comment