arXiv：cs.LG（机器学习，全量分类）· Jie Tang, Chuanlong Xie, Lixing Zhu·· 9 小时前AI 评分33
基于容忍度的公平性审计：违规认证与敏感性筛查
Tolerance-Based Fairness Auditing: Violation Certification and Sensitivity Screening
AI 导读
研究提出统一的基于容忍度公平性审计框架，面向违规认证与敏感性筛查两个互补目标。前者采用约束经验似然检验，结合最不利点校准与误报率控制，支持多子群同步审计；后者提出分裂经验似然及调整分裂经验似然检验，基于自适应边界代理原则用于早期预警。数值实验显示两类方法在错误控制与敏感性上存在不同权衡，并通过对 COMPAS 的分析展示了该框架在预测公平性审计中的应用。
正文
Abstract:As artificial intelligence is increasingly deployed, algorithmic unfairness has raised growing concerns and intensified demands for transparent fairness auditing. In practice, the tolerable degree of algorithmic unfairness depends on the specific legal, ethical, or application context. Given a prespecified tolerance threshold, an important statistical question is how to determine whether a group disparity exceeds the allowable tolerance across different auditing objectives. To address this problem, we develop a unified tolerance-based fairness auditing framework for two complementary auditing objectives: violation certification, which prioritizes control of false violation declarations, and sensitivity screening, which prioritizes reducing missed violations. For the first objective, we develop a constrained empirical likelihood test for formal settings that uses least-favorable-point calibration and can be combined with false flagging rate control for simultaneous subgroup auditing. For the second objective, we develop split empirical likelihood and adjusted split empirical likelihood tests using an adaptive boundary-proxy principle for early-warning settings. Numerical experiments show the distinct error-control--sensitivity trade-offs of these procedures. A COMPAS analysis illustrates the framework in predictive fairness auditing.
| Comments: | 47 pages, 8 figures |
| Subjects: | Machine Learning (stat.ML); Machine Learning (cs.LG); Methodology (stat.ME) |
| MSC classes: | 62G10 (Primary), 62G20, 68T05 (Secondary) |
| ACM classes: | I.2.6 |
| Cite as: | arXiv:2610.01005 [stat.ML] |
| (or arXiv:2610.01005v1 [stat.ML] for this version) | |
| https://doi.org/10.48550/arXiv.2610.01005 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Jie Tang [view email]
[v1]
Thu, 1 Oct 2026 03:49:53 UTC (1,273 KB)
来源：arXiv：cs.LG（机器学习，全量分类） · arxiv.org