ProFair: Proactive Fairness-Aware Learning in Semi-supervised Medical Image Classification

Abstract

Semi-Supervised Learning (SSL) has emerged as a critical enabler for data-efficient diagnostic tasks in medical imaging. However, existing SSL approaches generally neglect inherent inter-class learning difficulty disparities, introducing fairness limitations. Such models often tend to overfit to the class that is easy to classify, perform poorly when dealing with challenging samples (such as minority/complex classes), and fail to fully utilize the valuable knowledge contained in low-confidence samples. To bridge these gaps, we propose ProFair: a proactive fairness prototype-driven framework with adaptive perception for SSL. It provides fairer learning opportunities for samples with different difficulty levels. Specifically, the difficulty-aware adaptive soft label mechanism dynamically generates class-specific thresholds based on learning difficulty and optimizes label distribution through prototype similarity to model inter-class differences. The dual-path validated negative pseudo-label strategy employs dual-path noise-resistant verification with probability distribution and feature space. This strategy converts low-confidence samples into high-value training signals, thereby clearly reinforcing the decision boundary. The multi-objective global constraint loss integrates various constraint terms to jointly optimize diagnostic accuracy and model robustness. Extensive experiments on multiple public medical image datasets validate the effectiveness and superiority of ProFair. This research provides a theoretically rigorous and clinically verifiable solution for class-imbalanced medical data analysis.