Unsupervised Progressive Learning and the STAM Architecture

We first pose the Unsupervised Progressive Learning (UPL) problem: an online representation learning problem in which the learner observes a non-stationary and unlabeled data stream, and identifies a growing number of features that persist over time even though the data is not stored or replayed. To solve the UPL problem we propose the Self-Taught Associative Memory (STAM) architecture. Layered hierarchies of STAM modules learn based on a combination of online clustering, novelty detection, forgetting outliers, and storing only prototypical features rather than specific examples. We evaluate STAM representations using classification and clustering tasks. While there are no existing learning scenarios which are directly comparable to UPL, we compare the STAM architecture with two recent continual learning works; Memory Aware Synapses (MAS), and Gradient Episodic Memories (GEM), which have been modified to be suitable for the UPL setting.