Which deep learning layer to use for autoencoder using the deep learning designer?

I am trying to build a stacked autoencoder combined with one-class SVM for anomaly detection. I tried training each autoencoder and stack it together but I figured it would be easier if I do it in a deep learning architecture as I am facing some issue bringing the data after the dimension reduction.

Answer 1

[David Willingham]

Autoencoders can be created using a variety of architectures, so there is no single layer.

This page gives an overview, and examples for autoencoders in MATLAB.

[Ng Yong Jie]

Is it possible to create autoencoders layer using the deep network designer?