arXiv：cs.LG· Nithin Raghava Ramachandra Narla·· 3 小时前
FAPE 框架跨八领域评测后处理公平性约束：何时有效、何时有害
When Post-Processing Fairness Constraints Help and When They Harm: Evidence from Eight Cross-Domain Evaluations
AI 导读
研究提出 FAPE 四阶段框架，在刑事司法、收入预测、信贷、医疗、教育等八个领域评测 Fairlearn 的 ThresholdOptimizer 后处理干预。结果显示干预效果与基线差距相关：14 个高差距模型-领域对中 9 个差距改善，4 个接近公平的案例中 3 个反而恶化。部署时启动的 CUSUM 监控能在模拟偏移下区分从未达标与达标后回退的模型，单次部署审计不可靠。
正文
Abstract:Fairness audits in production ML typically occur once, at deployment, on a single domain. Both fail in practice: fairness can shift after retraining or a changing user base, and interventions validated on one dataset are rarely tested across the heterogeneous domains an organization deploys. We present FAPE (Fairness Auditing for Production Environments), a four-stage framework evaluating a single post-processing intervention, Fairlearn's ThresholdOptimizer, across eight domain evaluations: criminal justice, income prediction, legal admissions, credit lending, agricultural lending, a multi-domain benchmark corpus, healthcare, and education. Each is scored on demographic parity and equalized odds difference, plus disparate impact ratio and accuracy cost where computable. Intervention effectiveness tracks baseline disparity magnitude: across model-domain pairs the constraint improved disparity in 9 of 14 high-disparity cases and worsened it in 3 of 4 near-fair ones. Each of the five high-disparity exceptions reverses under one of two measurement checks, a minimum group size or thresholds fit on held-out data. A CUSUM monitor started at deployment, tested on a simulated shift, separates constrained models that never met a 0.1 parity convention from those that met it and later regressed. A single deployment-time audit is therefore an unreliable guide, which argues for baseline-disparity screening and continuous monitoring
| Comments: | 18 pages, 6 figures, 2 tables. Code, data loaders and figures: this http URL |
| Subjects: | Machine Learning (cs.LG); Computers and Society (cs.CY) |
| Cite as: | arXiv:2609.26955 [cs.LG] |
| (or arXiv:2609.26955v3 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2609.26955 arXiv-issued DOI via DataCite |
Submission history
From: Nithin Raghava Ramachandra Narla [view email]
[v1]
Tue, 22 Sep 2026 18:46:00 UTC (1,657 KB)
[v2]
Tue, 29 Sep 2026 07:09:28 UTC (1,657 KB)
[v3]
Wed, 7 Oct 2026 21:33:20 UTC (1,647 KB)
来源：arXiv：cs.LG · arxiv.org