Performance difference between binary_crossentropy vs categorical_cross_entropy
Is there a difference between binary_crossentropy vs categorical_cross_entropy in terms of speed? I use Keras with the TensorFlow backend.
Is there a difference between binary_crossentropy vs categorical_cross_entropy in terms of speed? I use Keras with the TensorFlow backend.
There is a difference, but it is too small and is not noticeable. If you have 2 classes, there is no need to use categorical, just binary cross-entropy is enough.