Open U6C opened 7 years ago
If you want to really do that, then you can fix the layers except the one you want to train: https://keras.io/getting-started/faq/#how-can-i-freeze-keras-layers . A guide in tensorflow might be http://cmgreen.io/2016/01/04/tensorflow_deep_autoencoder.html , it should not bee too hard to adapt,
Correct me if I am wrong, the current implementation does not seem to be trained layer-by-layer. I've been searching for examples for stacked autoencoders trained in layerwise fashion using Keras. Any thoughts? Thank you.