Weakly Supervised Contrastive Learning

TL;DR

This paper introduces Weakly Supervised Contrastive Learning (WCL), a novel framework that enhances unsupervised visual representation by combining instance discrimination with weak label supervision, leading to state-of-the-art semi-supervised results.

Contribution

The paper proposes a dual-head contrastive learning framework that incorporates graph-based weak labels and a multi-crop strategy to improve representation quality and semi-supervised performance.

Findings

WCL achieves new state-of-the-art results in semi-supervised learning.

WCL outperforms existing methods with only 1 ext{-}10 ext{\%} labeled data.

WCL improves representation quality across various datasets.

Abstract

Unsupervised visual representation learning has gained much attention from the computer vision community because of the recent achievement of contrastive learning. Most of the existing contrastive learning frameworks adopt the instance discrimination as the pretext task, which treating every single instance as a different class. However, such method will inevitably cause class collision problems, which hurts the quality of the learned representation. Motivated by this observation, we introduced a weakly supervised contrastive learning framework (WCL) to tackle this issue. Specifically, our proposed framework is based on two projection heads, one of which will perform the regular instance discrimination task. The other head will use a graph-based method to explore similar samples and generate a weak label, then perform a supervised contrastive learning task based on the weak label to pull the similar images closer. We further introduced a K-Nearest Neighbor based multi-crop strategy to expand the number of positive samples. Extensive experimental results demonstrate WCL improves the quality of self-supervised representations across different datasets. Notably, we get a new state-of-the-art result for semi-supervised learning. With only 1\% and 10\% labeled examples, WCL achieves 65\% and 72\% ImageNet Top-1 Accuracy using ResNet50, which is even higher than SimCLRv2 with ResNet101.