A Mutual Information Perspective on Federated Contrastive Learning

TL;DR

This paper explores federated contrastive learning through mutual information, proposing extensions for semi-supervised settings and analyzing the impact of data heterogeneity on performance.

Contribution

It introduces a mutual information perspective to federated contrastive learning, extending SimCLR for semi-supervised scenarios and analyzing non-i.i.d. data effects.

Findings

Adding user verification improves global mutual information bounds.

Supervised SimCLR can be achieved with label-based contrastive loss and auxiliary classification.

Global mutual information maximization benefits some non-i.i.d. data sources but not others.

Abstract

We investigate contrastive learning in the federated setting through the lens of SimCLR and multi-view mutual information maximization. In doing so, we uncover a connection between contrastive representation learning and user verification; by adding a user verification loss to each client's local SimCLR loss we recover a lower bound to the global multi-view mutual information. To accommodate for the case of when some labelled data are available at the clients, we extend our SimCLR variant to the federated semi-supervised setting. We see that a supervised SimCLR objective can be obtained with two changes: a) the contrastive loss is computed between datapoints that share the same label and b) we require an additional auxiliary head that predicts the correct labels from either of the two views. Along with the proposed SimCLR extensions, we also study how different sources of non-i.i.d.-ness can impact the performance of federated unsupervised learning through global mutual information maximization; we find that a global objective is beneficial for some sources of non-i.i.d.-ness but can be detrimental for others. We empirically evaluate our proposed extensions in various tasks to validate our claims and furthermore demonstrate that our proposed modifications generalize to other pretraining methods.