Adversarial Contrastive Self-Supervised Learning

TL;DR

This paper introduces a novel self-supervised learning method using online hard negative mining within a student-teacher network, significantly improving image representation learning efficiency.

Contribution

It proposes a new self-supervised framework that incorporates hard negative mining with a student-teacher architecture, enhancing learning from unlabeled data.

Findings

Effective on ILSVRC-2012 dataset

Outperforms existing self-supervised methods

Improves label efficiency in training

Abstract

Recently, learning from vast unlabeled data, especially self-supervised learning, has been emerging and attracted widespread attention. Self-supervised learning followed by the supervised fine-tuning on a few labeled examples can significantly improve label efficiency and outperform standard supervised training using fully annotated data. In this work, we present a novel self-supervised deep learning paradigm based on online hard negative pair mining. Specifically, we design a student-teacher network to generate multi-view of the data for self-supervised learning and integrate hard negative pair mining into the training. Then we derive a new triplet-like loss considering both positive sample pairs and mined hard negative sample pairs. Extensive experiments demonstrate the effectiveness of the proposed method and its components on ILSVRC-2012.