How can I create and implement Local response Normalization using tensor ops and implement it in my tf.keras models. And should I use it in modern CNN’s? why or why not?
Expression for the Norm :-
You can this a Keras lambda layer:
https://www.tensorflow.org/api_docs/python/tf/nn/local_response_normalization
It was requested in Keras many times, I think the last one was:
1 Like