neural network - What is the impact of the size of the hidden layer in Autoencoders

Question

I am revisiting autoencoders recently and I found most of the discussions are on its relation to PCA. But there is little discussion on the size of the hidden layers.

Are there any conclusions on the impact of the size of the hidden layer in autoencoders other than conventional hyperparameter searching?

question from:https://stackoverflow.com/questions/65909366/what-is-the-impact-of-the-size-of-the-hidden-layer-in-autoencoders

Answer 1

Waitting for answers