Whitening for Self-Supervised Representation Learning

TL;DR

This paper introduces a novel self-supervised learning method using whitening of latent features, eliminating the need for negative samples and simplifying the training process while preventing feature collapse.

Contribution

It proposes a new whitening-based loss for SSL that avoids negatives, simplifies architecture, and allows multiple positives from the same image.

Findings

The whitening method prevents feature collapse.

No negatives are required for training.

Multiple positives can be extracted from a single image.

Abstract

Most of the current self-supervised representation learning (SSL) methods are based on the contrastive loss and the instance-discrimination task, where augmented versions of the same image instance ("positives") are contrasted with instances extracted from other images ("negatives"). For the learning to be effective, many negatives should be compared with a positive pair, which is computationally demanding. In this paper, we propose a different direction and a new loss function for SSL, which is based on the whitening of the latent-space features. The whitening operation has a "scattering" effect on the batch samples, avoiding degenerate solutions where all the sample representations collapse to a single point. Our solution does not require asymmetric networks and it is conceptually simple. Moreover, since negatives are not needed, we can extract multiple positive pairs from the same image instance. The source code of the method and of all the experiments is available at: https://github.com/htdt/self-supervised.