Learning Time-Varying Convexifications of Multiple Fairness Measures

2025-08-19

AI Summary

This paper addresses online fair learning under graph-structured feedback constraints: specifically, how to dynamically learn time-varying regularization coefficients for fairness preferences—defined as convex combinations of multiple fairness metrics—whose weights evolve over time and are unknown a priori. To this end, we propose the first dynamic convex combination learning framework for multi-objective fairness, modeling time-varying weights as online optimization variables. Our method jointly incorporates graph-structured feedback and adaptive convex optimization to co-model group- and individual-level fairness. Unlike conventional static-weight approaches, it enables rapid tracking of fairness preference drifts even under sparse and delayed graph feedback. Experiments demonstrate that our framework significantly improves both adaptation speed and stability in dynamic environments, offering a scalable theoretical foundation and practical methodology for online fair decision-making.

Technology Category

• Machine Learning: Online Learning & Bandits
• Search and Optimization: Learning to Search
• Constraint Satisfaction and Optimization: Constraint Learning and Acquisition

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• 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

Abstract

There is an increasing appreciation that one may need to consider multiple measures of fairness, e.g., considering multiple group and individual fairness notions. The relative weights of the fairness regularisers are a priori unknown, may be time varying, and need to be learned on the fly. We consider the learning of time-varying convexifications of multiple fairness measures with limited graph-structured feedback.

Problem

• Learning time-varying convexifications of multiple fairness measures
• Determining unknown and changing fairness regulariser weights
• Addressing limited graph-structured feedback in fairness optimization

Innovation

• Learning time-varying convexifications of fairness measures
• Multiple group and individual fairness notions combined
• Graph-structured feedback for adaptive weight optimization