On Mutual Information Maximization for Representation Learning

Michael Tschannen, Josip Djolonga, Paul K. Rubenstein, Sylvain Gelly, Mario Lucic

Abstract: Many recent methods for unsupervised or self-supervised representation learning train feature extractors by maximizing an estimate of the mutual information (MI) between different views of the data. This comes with several immediate problems: For example, MI is notoriously hard to estimate, and using it as an objective for representation learning may lead to highly entangled representations due to its invariance under arbitrary invertible transformations. Nevertheless, these methods have been repeatedly shown to excel in practice. In this paper we argue, and provide empirical evidence, that the success of these methods cannot be attributed to the properties of MI alone, and that they strongly depend on the inductive bias in both the choice of feature extractor architectures and the parametrization of the employed MI estimators. Finally, we establish a connection to deep metric learning and argue that this interpretation may be a plausible explanation for the success of the recently introduced methods.

Code: https://storage.googleapis.com/mi_for_rl_files/code.zip

Keywords: mutual information, representation learning, unsupervised learning, self-supervised learning

TL;DR: The success of recent mutual information (MI)-based representation learning approaches strongly depends on the inductive bias in both the choice of network architectures and the parametrization of the employed MI estimators.

Community Implementations: [ 2 code implementations](https://www.catalyzex.com/paper/arxiv:1907.13625/code)