Can contrastive learning avoid shortcut solutions?

TL;DR

This paper investigates how contrastive learning can inadvertently focus on shortcuts, proposes a method called implicit feature modification to encourage capturing diverse features, and demonstrates improved performance on vision and medical imaging tasks.

Contribution

It introduces implicit feature modification (IFM), a novel technique to guide contrastive learning towards capturing a broader set of predictive features.

Findings

IFM reduces feature suppression in contrastive learning

Improved downstream task performance with IFM

Effective on vision and medical imaging datasets

Abstract

The generalization of representations learned via contrastive learning depends crucially on what features of the data are extracted. However, we observe that the contrastive loss does not always sufficiently guide which features are extracted, a behavior that can negatively impact the performance on downstream tasks via "shortcuts", i.e., by inadvertently suppressing important predictive features. We find that feature extraction is influenced by the difficulty of the so-called instance discrimination task (i.e., the task of discriminating pairs of similar points from pairs of dissimilar ones). Although harder pairs improve the representation of some features, the improvement comes at the cost of suppressing previously well represented features. In response, we propose implicit feature modification (IFM), a method for altering positive and negative samples in order to guide contrastive models towards capturing a wider variety of predictive features. Empirically, we observe that IFM reduces feature suppression, and as a result improves performance on vision and medical imaging tasks. The code is available at: \url{https://github.com/joshr17/IFM}.