Dual Temperature Helps Contrastive Learning Without Many Negative Samples: Towards Understanding and Simplifying MoCo

TL;DR

This paper introduces a simplified contrastive learning framework using dual temperature scaling, removing the need for large dictionaries, and demonstrates improved performance and a unified understanding of SSL methods.

Contribution

It proposes a dual temperature approach to simplify MoCo, removing the dictionary and momentum, and bridges the understanding between contrastive and non-contrastive frameworks.

Findings

SimMoCo and SimCo outperform MoCo v2 in experiments.

Dual temperature effectively replaces large dictionaries in contrastive learning.

The work unifies contrastive and non-contrastive SSL frameworks.

Abstract

Contrastive learning (CL) is widely known to require many negative samples, 65536 in MoCo for instance, for which the performance of a dictionary-free framework is often inferior because the negative sample size (NSS) is limited by its mini-batch size (MBS). To decouple the NSS from the MBS, a dynamic dictionary has been adopted in a large volume of CL frameworks, among which arguably the most popular one is MoCo family. In essence, MoCo adopts a momentum-based queue dictionary, for which we perform a fine-grained analysis of its size and consistency. We point out that InfoNCE loss used in MoCo implicitly attract anchors to their corresponding positive sample with various strength of penalties and identify such inter-anchor hardness-awareness property as a major reason for the necessity of a large dictionary. Our findings motivate us to simplify MoCo v2 via the removal of its dictionary as well as momentum. Based on an InfoNCE with the proposed dual temperature, our simplified frameworks, SimMoCo and SimCo, outperform MoCo v2 by a visible margin. Moreover, our work bridges the gap between CL and non-CL frameworks, contributing to a more unified understanding of these two mainstream frameworks in SSL. Code is available at: https://bit.ly/3LkQbaT.