Learning Label Refinement and Threshold Adjustment for Imbalanced Semi-Supervised Learning

TL;DR

This paper introduces SEVAL, a novel pseudo-label refinement and threshold adjustment method for imbalanced semi-supervised learning, improving pseudo-label quality and outperforming existing SSL techniques.

Contribution

The paper proposes SEVAL, a class-balanced pseudo-label optimization approach that enhances pseudo-label accuracy and robustness in imbalanced SSL scenarios.

Findings

SEVAL outperforms state-of-the-art SSL methods in imbalanced settings.

SEVAL improves pseudo-label accuracy and class-wise correctness.

The method is simple, flexible, and applicable to various SSL techniques.

Abstract

Semi-supervised learning (SSL) algorithms struggle to perform well when exposed to imbalanced training data. In this scenario, the generated pseudo-labels can exhibit a bias towards the majority class, and models that employ these pseudo-labels can further amplify this bias. Here we investigate pseudo-labeling strategies for imbalanced SSL including pseudo-label refinement and threshold adjustment, through the lens of statistical analysis. We find that existing SSL algorithms which generate pseudo-labels using heuristic strategies or uncalibrated model confidence are unreliable when imbalanced class distributions bias pseudo-labels. To address this, we introduce SEmi-supervised learning with pseudo-label optimization based on VALidation data (SEVAL) to enhance the quality of pseudo-labelling for imbalanced SSL. We propose to learn refinement and thresholding parameters from a partition of the training dataset in a class-balanced way. SEVAL adapts to specific tasks with improved pseudo-labels accuracy and ensures pseudo-labels correctness on a per-class basis. Our experiments show that SEVAL surpasses state-of-the-art SSL methods, delivering more accurate and effective pseudo-labels in various imbalanced SSL situations. SEVAL, with its simplicity and flexibility, can enhance various SSL techniques effectively. The code is publicly available (https://github.com/ZerojumpLine/SEVAL).