paperbot · PL 论文追踪

全部论文 待读 收藏

P2 Approximation-guided Fairness Testing through Discriminatory Space Analysis

通过歧视空间分析的近似引导公平性测试
ASE Proceedings of the 39th IEEE/ACM International Conference on Automated Software Engineering2024匹配度 3/10 · 该论文属于 ML 系统的公平性测试（个体公平性黑盒测试），与画像关注的「在 ML 系统中应用静态分析」仅边缘相关——它是测试而非静态分析，且完全未涉及 LLM/Agent、浮点程序分析或贝叶斯程序分析。其结论主要靠决策树近似+路径采样的工程启发式在效率指标上超越 ExpGA，缺乏画像偏好的理论形式化或问题失效机理分析。引用 6 inbox
Zhenjiang Zhao, Takahisa Toda, Takashi Kitamura
测试、调试与修复机器学习与程序语言
crossref: 2026-09-12dblp: 2026-09-12

AI 速览

一句话：提出黑盒个体公平性测试算法 Aft，用决策树近似模型采样路径生成歧视实例。

随着机器学习系统在高社会影响领域广泛应用，其公平性问题日益突出。个体公平性测试旨在寻找违反个体公平性的个体歧视实例（IDI），但现有黑盒方法在效率和实例多样性上仍有不足。本文提出 Aft（Approximation-guided Fairness Testing），一种通过歧视空间分析实现的黑盒测试算法。Aft 基于决策树构建待测分类器的近似模型，并沿决策树路径采样生成测试用例，以搜索潜在 IDI。实验表明，与当前最优黑盒个体公平性测试算法 ExpGA 相比，Aft 在效率上提升 3.42 倍，所识别 IDI 的多样性提升 1.16 倍。

问题
如何高效地为黑盒 ML 分类器进行个体公平性测试，发现违反个体公平性的歧视实例。
方法
Aft 基于决策树构造近似模型，并通过采样决策树路径生成测试用例，以引导歧视空间搜索。
评估
与当前最优黑盒个体公平性测试算法 ExpGA 进行实验对比，效率提升 3.42 倍，识别出的 IDI 多样性提升 1.16 倍。
原文摘要（Abstract）

As machine learning (ML) systems are increasingly used in various fields, including tasks with high social impact, concerns about their fairness are growing. To address these concerns, individual fairness testing (IFT) has been introduced to identify individual discriminatory instances (IDIs) that indicate the violation of individual fairness in a given ML classifier. In this paper, we propose a black-box testing algorithm for IFT, named Aft (short for Approximation-guided Fairness Testing). Aft constructs approximate models based on decision trees, and generates test cases by sampling paths of the decision trees. Our evaluation by experiments confirms that Aft outperforms the state-of-the-art black-box IFT algorithm ExpGA both in efficiency (by 3.42 times) and diversity of IDIs identified by algorithms (by 1.16 times).CCS CONCEPTS• Software and its engineering → Software testing and debugging.

链接与引用

DOI 原文 · PDF（开放获取） · DBLP

BibTeX 
@inproceedings{ZhaoT024,
  title = {Approximation-guided Fairness Testing through Discriminatory Space Analysis},
  author = {Zhenjiang Zhao and Takahisa Toda and Takashi Kitamura},
  booktitle = {ASE},
  year = {2024},
  doi = {10.1145/3691620.3695481}
}