Fair Classification by Direct Intervention on Operating Characteristics

2025-09-29

AI Summary

This paper addresses the challenge of simultaneously satisfying multiple group fairness criteria—such as demographic parity, equal opportunity, and predictive equality—in binary classification. We propose a post-processing fairification method that leverages the grouped ROC convex hulls of a base classifier to characterize decision boundaries across subpopulations. Our core innovation is a randomized threshold-mixing rule that minimizes the number of interventions to the original model’s predictions while satisfying multiple linear-fractional fairness constraints across sensitive attributes. The method requires no retraining and supports complex, composite fairness requirements. Evaluated on COMPAS and ACSIncome datasets, it achieves near-optimal fairness across multiple metrics with minimal prediction adjustments, attaining accuracy close to the oracle upper bound. Empirically, it outperforms existing post-processing and ensemble approaches in overall fairness-accuracy trade-off.

Technology Category

• Machine Learning: Ethics, Bias, and Fairness
• Computer Vision: Bias, Fairness & Privacy
• Game Theory and Economic Paradigms: Fair Division

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Economics, Online Markets and Human Computation: Fairness, privacy, and diversity in economic environments
• Social Networks and Social Media: Fairness and bias in social network and social media analysis

Abstract

We develop new classifiers under group fairness in the attribute-aware setting for binary classification with multiple group fairness constraints (e.g., demographic parity (DP), equalized odds (EO), and predictive parity (PP)). We propose a novel approach, applicable to linear fractional constraints, based on directly intervening on the operating characteristics of a pre-trained base classifier, by (i) identifying optimal operating characteristics using the base classifier's group-wise ROC convex hulls and (ii) post-processing the base classifier to match those targets. As practical post-processors, we consider randomizing a mixture of group-wise thresholding rules subject to minimizing the expected number of interventions. We further extend our approach to handle multiple protected attributes and multiple linear fractional constraints. On standard datasets (COMPAS and ACSIncome), our methods simultaneously satisfy approximate DP, EO, and PP with few interventions and a near-oracle drop in accuracy; comparing favorably to previous methods.

Problem

• Develop classifiers ensuring group fairness with multiple constraints
• Directly intervene on operating characteristics of pre-trained classifiers
• Handle multiple protected attributes and fairness constraints simultaneously

Innovation

• Direct intervention on classifier operating characteristics
• Using group-wise ROC convex hulls optimization
• Post-processing with randomized thresholding rules