Contrastive Learning with Hard Negative Samples

TL;DR

This paper introduces an unsupervised sampling method for contrastive learning that effectively selects hard negatives, improving representation quality without additional computational cost.

Contribution

It proposes a novel family of unsupervised hard negative sampling techniques that enhance contrastive learning performance across various modalities.

Findings

Improved downstream task performance

Tightly clustered class representations

No additional computational overhead

Abstract

How can you sample good negative examples for contrastive learning? We argue that, as with metric learning, contrastive learning of representations benefits from hard negative samples (i.e., points that are difficult to distinguish from an anchor point). The key challenge toward using hard negatives is that contrastive methods must remain unsupervised, making it infeasible to adopt existing negative sampling strategies that use true similarity information. In response, we develop a new family of unsupervised sampling methods for selecting hard negative samples where the user can control the hardness. A limiting case of this sampling results in a representation that tightly clusters each class, and pushes different classes as far apart as possible. The proposed method improves downstream performance across multiple modalities, requires only few additional lines of code to implement, and introduces no computational overhead.