Semantic Positive Pairs for Enhancing Visual Representation Learning of Instance Discrimination Methods

TL;DR

This paper introduces a method to identify and utilize semantically similar positive pairs in self-supervised learning, improving visual representations by capturing more meaningful features beyond traditional data augmentation.

Contribution

The proposed approach enhances instance discrimination SSL methods by incorporating semantic positive pairs, leading to better representations across multiple datasets and tasks.

Findings

Consistent performance improvements over baseline methods.

4.1% accuracy gain on ImageNet with MoCo-v2.

Effective in semi-supervised, transfer learning, and object detection tasks.

Abstract

Self-supervised learning algorithms (SSL) based on instance discrimination have shown promising results, performing competitively or even outperforming supervised learning counterparts in some downstream tasks. Such approaches employ data augmentation to create two views of the same instance (i.e., positive pairs) and encourage the model to learn good representations by attracting these views closer in the embedding space without collapsing to the trivial solution. However, data augmentation is limited in representing positive pairs, and the repulsion process between the instances during contrastive learning may discard important features for instances that have similar categories. To address this issue, we propose an approach to identify those images with similar semantic content and treat them as positive instances, thereby reducing the chance of discarding important features during representation learning and increasing the richness of the latent representation. Our approach is generic and could work with any self-supervised instance discrimination frameworks such as MoCo and SimSiam. To evaluate our method, we run experiments on three benchmark datasets: ImageNet, STL-10 and CIFAR-10 with different instance discrimination SSL approaches. The experimental results show that our approach consistently outperforms the baseline methods across all three datasets; for instance, we improve upon the vanilla MoCo-v2 by 4.1% on ImageNet under a linear evaluation protocol over 800 epochs. We also report results on semi-supervised learning, transfer learning on downstream tasks, and object detection.