Disentangling and Learning Robust Representations with Natural Clustering

Javier Antoran, Antonio Miguel

Learning representations that disentangle the underlying factors of variability in data is an intuitive way to achieve generalization in deep models. In this work, we address the s…