I've been working on Feature Learning related projects. Recently, I came across the research paper "Representation Learning: A Unified Deep Learning Framework for Automatic Prostate MR Segmentation" that uses a stacked Independent Subspace analysis (ISA) approach for unsupervised Feature learning. I decided to experiment on it. So far I've been able to implement the basic ISA algorithm in Python. I'm now trying to implement the hierarchical (Stacked) Network that use ISA and PCA as sub units. Basically I want to create a network in which the second layer convolves over the output of the first layer. Can somebody guide me on how i could implement this ( just like in the research paper). Or if there are any resources/similar examples that I can refer to?
Link to paper - Conference Paper Representation Learning: A Unified Deep Learning Framework f...