Modular Autoencoders for Ensemble Feature Extraction

Henry W J Reeve, Gavin Brown

We introduce the concept of a Modular Autoencoder (MAE), capable of learning a set of diverse but complementary representations from unlabelled data, that can later be used for sup…