Incremental False Negative Detection for Contrastive Learning

TL;DR

This paper introduces an incremental false negative detection framework for contrastive learning that dynamically identifies and removes false negatives, improving self-supervised learning performance on large-scale datasets.

Contribution

It proposes a novel method to detect and remove false negatives during contrastive learning, addressing a key limitation in existing self-supervised approaches.

Findings

Outperforms existing methods on multiple benchmarks

Effectively detects high-quality false negatives during training

Improves learning performance in resource-limited settings

Abstract

Self-supervised learning has recently shown great potential in vision tasks through contrastive learning, which aims to discriminate each image, or instance, in the dataset. However, such instance-level learning ignores the semantic relationship among instances and sometimes undesirably repels the anchor from the semantically similar samples, termed as "false negatives". In this work, we show that the unfavorable effect from false negatives is more significant for the large-scale datasets with more semantic concepts. To address the issue, we propose a novel self-supervised contrastive learning framework that incrementally detects and explicitly removes the false negative samples. Specifically, following the training process, our method dynamically detects increasing high-quality false negatives considering that the encoder gradually improves and the embedding space becomes more semantically structural. Next, we discuss two strategies to explicitly remove the detected false negatives during contrastive learning. Extensive experiments show that our framework outperforms other self-supervised contrastive learning methods on multiple benchmarks in a limited resource setup.