The dynamics of representation learning in shallow, non-linear autoencoders

Open in new window