OxEnsemble: Fair Ensembles for Low-Data Classification

2025-12-10

AI Summary

Addressing fair classification in low-data, highly imbalanced medical imaging domains—where false negatives incur high clinical costs—remains challenging due to scarce labeled samples and stringent fairness requirements. Method: We propose a lightweight model ensemble framework that enforces individual fairness constraints. It introduces a novel member-prediction aggregation paradigm under fairness constraints, a data-efficient training mechanism leveraging reusable validation sets, and the first theoretical convergence guarantee for fair ensembling under low-data regimes. Contributions/Results: Evaluated on multiple medical imaging datasets, our method reduces the equal opportunity difference by 37% on average while maintaining >98% accuracy. Training overhead is only marginally higher than single-model fine-tuning, significantly improving the fairness–accuracy trade-off without compromising computational efficiency or clinical utility.

Technology Category

• Machine Learning: Ensemble Methods
• Computer Vision: Bias, Fairness & Privacy
• Philosophy and Ethics of AI: Bias, Fairness & Equity

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Social Networks and Social Media: Fairness and bias in social network and social media analysis

Abstract

We address the problem of fair classification in settings where data is scarce and unbalanced across demographic groups. Such low-data regimes are common in domains like medical imaging, where false negatives can have fatal consequences. We propose a novel approach emph{OxEnsemble} for efficiently training ensembles and enforcing fairness in these low-data regimes. Unlike other approaches, we aggregate predictions across ensemble members, each trained to satisfy fairness constraints. By construction, emph{OxEnsemble} is both data-efficient, carefully reusing held-out data to enforce fairness reliably, and compute-efficient, requiring little more compute than used to fine-tune or evaluate an existing model. We validate this approach with new theoretical guarantees. Experimentally, our approach yields more consistent outcomes and stronger fairness-accuracy trade-offs than existing methods across multiple challenging medical imaging classification datasets.

Problem

• Addresses fair classification in low-data, unbalanced demographic settings
• Proposes OxEnsemble for efficient fairness enforcement with limited data
• Improves fairness-accuracy trade-offs in medical imaging classification tasks

Innovation

• Ensemble training with fairness constraints
• Data-efficient reuse of held-out data
• Compute-efficient fine-tuning of existing models