Leveraging Label Proportion Prior for Class-Imbalanced Semi-Supervised Learning

TL;DR

This paper introduces a novel Proportion Loss regularization for semi-supervised learning that addresses class imbalance by aligning predictions with global class distributions, improving minority class performance.

Contribution

It is the first to incorporate Proportion Loss from learning from label proportions into SSL, enhancing bias mitigation in imbalanced datasets.

Findings

Improves performance of FixMatch and ReMixMatch on long-tailed CIFAR-10.

Achieves superior results under scarce-label conditions.

Stabilizes training with a stochastic variant of Proportion Loss.

Abstract

Semi-supervised learning (SSL) often suffers under class imbalance, where pseudo-labeling amplifies majority bias and suppresses minority performance. We address this issue with a lightweight framework that, to our knowledge, is the first to introduce Proportion Loss from learning from label proportions (LLP) into SSL as a regularization term. Proportion Loss aligns model predictions with the global class distribution, mitigating bias across both majority and minority classes. To further stabilize training, we formulate a stochastic variant that accounts for fluctuations in mini-batch composition. Experiments on the Long-tailed CIFAR-10 benchmark show that integrating Proportion Loss into FixMatch and ReMixMatch consistently improves performance over the baselines across imbalance severities and label ratios, and achieves competitive or superior results compared to existing CISSL methods, particularly under scarce-label conditions.