Improving Semi-Supervised Contrastive Learning via Entropy-Weighted Confidence Integration of Anchor-Positive Pairs

TL;DR

This paper introduces an entropy-based confidence weighting method for semi-supervised contrastive learning, allowing more samples to be used effectively and improving classification accuracy especially with limited labeled data.

Contribution

It proposes a novel loss function that estimates sample confidence via entropy and adaptively weights samples, enhancing semi-supervised contrastive learning.

Findings

Improves classification accuracy under low-label conditions.

Achieves more stable learning performance.

Enables pseudo-labeling for previously excluded samples.

Abstract

Conventional semi-supervised contrastive learning methods assign pseudo-labels only to samples whose highest predicted class probability exceeds a predefined threshold, and then perform supervised contrastive learning using those selected samples. In this study, we propose a novel loss function that estimates the confidence of each sample based on the entropy of its predicted probability distribution and applies confidence-based adaptive weighting. This approach enables pseudo-label assignment even to samples that were previously excluded from training and facilitates contrastive learning that accounts for the confidence of both anchor and positive samples in a more principled manner. Experimental results demonstrate that the proposed method improves classification accuracy and achieves more stable learning performance even under low-label conditions.