Post-Training Fairness Control: A Single-Train Framework for Dynamic Fairness in Recommendation

2026-01-28

AI Summary

This work proposes Cofair, a framework that addresses the inflexibility of existing fairness-aware recommendation systems, which typically hardcode fairness constraints during training and cannot adapt to diverse post-hoc fairness requirements. Cofair employs a shared representation layer coupled with fairness-conditioned adapters, enabling dynamic adjustment of recommendation fairness levels after a single training run. It further introduces a user-level monotonicity regularizer to guarantee that fairness strictly improves as the target fairness constraint becomes more stringent. Theoretical analysis demonstrates that its adversarial training mechanism effectively bounds demographic parity violation. Extensive experiments show that Cofair is compatible with various backbone models and achieves competitive or superior fairness-accuracy trade-offs across multiple datasets, while eliminating the need for repeated training under different fairness objectives.

Technology Category

• Machine Learning: Ethics, Bias, and Fairness
• Computer Vision: Bias, Fairness & Privacy
• Multiagent Systems: Adversarial Agents

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Economics, Online Markets and Human Computation: Economics and fairness of platforms and recommendation systems
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Despite growing efforts to mitigate unfairness in recommender systems, existing fairness-aware methods typically fix the fairness requirement at training time and provide limited post-training flexibility. However, in real-world scenarios, diverse stakeholders may demand differing fairness requirements over time, so retraining for different fairness requirements becomes prohibitive. To address this limitation, we propose Cofair, a single-train framework that enables post-training fairness control in recommendation. Specifically, Cofair introduces a shared representation layer with fairness-conditioned adapter modules to produce user embeddings specialized for varied fairness levels, along with a user-level regularization term that guarantees user-wise monotonic fairness improvements across these levels. We theoretically establish that the adversarial objective of Cofair upper bounds demographic parity and the regularization term enforces progressive fairness at user level. Comprehensive experiments on multiple datasets and backbone models demonstrate that our framework provides dynamic fairness at different levels, delivering comparable or better fairness-accuracy curves than state-of-the-art baselines, without the need to retrain for each new fairness requirement. Our code is publicly available at https://github.com/weixinchen98/Cofair.

Problem

• fairness control
• post-training
• recommender systems
• dynamic fairness
• fairness requirements

Innovation

• post-training fairness control
• fairness-conditioned adapter
• dynamic fairness
• user-level regularization
• single-train framework