Self-Supervised Representation Learning on Manifolds

Abstract

We explore the use of a topological manifold, represented as a collection of charts, as the target space of neural network based representation learning tasks. This is achieved by a simple adjustment to the output of an encoder's network architecture plus the addition of a maximal mean discrepancy based loss function for regularization. Most algorithms in representation learning are easily adaptable to our framework and we demonstrate its effectiveness by adjusting SimCLR to have a manifold encoding space. Our experiments show that we obtain a substantial performance boost over the baseline for low dimensional encodings. Code for reproducing experiments is provided at https://github.com/ekorman/neurve.