Loading...
Fairness-Aware Meta-Learning via Nash Bargaining
Zeng, Yi Yang, Xuelin Chen, Li Ferrer, Cristian Jin, Ming Jordan, Michael Jia, Ruoxi
Zeng, Yi
Yang, Xuelin
Chen, Li
Ferrer, Cristian
Jin, Ming
Jordan, Michael
Jia, Ruoxi
Supervisor
Department
Machine Learning
Embargo End Date
Type
Conference proceeding
Date
2024
License
Language
English
Collections
Research Projects
Organizational Units
Journal Issue
Abstract
To address issues of group-level fairness in machine learning, it is natural to adjust model parameters based on specific fairness objectives over a sensitive-attributed validation set. Such an adjustment procedure can be cast within a meta-learning framework. However, naive integration of fairness goals via meta-learning can cause hypergradient conflicts for subgroups, resulting in unstable convergence and compromising model performance and fairness. To navigate this issue, we frame the resolution of hypergradient conflicts as a multi-player cooperative bargaining game. We introduce a two-stage meta-learning framework in which the first stage involves the use of a Nash Bargaining Solution (NBS) to resolve hypergradient conflicts and steer the model toward the Pareto front, and the second stage optimizes with respect to specific fairness goals.Our method is supported by theoretical results, notably a proof of the NBS for gradient aggregation free from linear independence assumptions, a proof of Pareto improvement, and a proof of monotonic improvement in validation loss. We also show empirical effects across various fairness objectives in six key fairness datasets and two image classification tasks.
Citation
Y. Zeng et al., “Fairness-Aware Meta-Learning via Nash Bargaining,” Adv Neural Inf Process Syst, vol. 37, pp. 83235–83267, Dec. 2024.
Source
Advances in Neural Information Processing Systems (NeurIPS 2024)
Conference
Keywords
Fairness-aware meta-learning, Nash Bargaining Solution (NBS), Hypergradient conflicts, Bi-level optimization, Group-level fairness
Subjects
Source
Publisher
NEURIPS