Database Repair Meets Algorithmic Fairness

Database Repair Meets Algorithmic Fairness
复制标题

数据库修复满足算法公平性

DOI:
10.1145/3422648.3422657
复制
发表时间:
2020
期刊:
ACM SIGMOD Record
影响因子:
--
通讯作者:
Dan Suciu
Dan Suciu
中科院分区:
--
文献类型:
--
作者:
Babak Salimi;Bill Howe;Dan Suciu

文献摘要

参考文献

被引文献

相似文献

公平性越来越被认为是机器学习系统的关键组成部分。然而,这些系统训练的基础数据往往反映了歧视,这表明数据库修复问题。现有的公平性处理方法依赖于统计相关性,这些相关性可能会被异常现象所愚弄,比如辛普森悖论。基于因果关系的公平定义的建议可以正确地模拟其中一些情况,但它们依赖于基本因果模型的背景知识。在本文中,我们正式的情况下,作为一个数据库修复问题,证明公平的分类器的充分条件,而不是一个完整的因果模型的容许变量。我们表明,这些条件正确地捕捉微妙的违反公平。然后,我们使用这些条件作为数据库修复算法的基础,这些算法为在其训练标签上训练的分类器提供可证明的公平性保证。我们证明了我们提出的技术与实验结果的有效性。
Fairness is increasingly recognized as a critical component of machine learning systems. However, it is the underlying data on which these systems are trained that often reflect discrimination, suggesting a database repair problem. Existing treatments of fairness rely on statistical correlations that can be fooled by anomalies, such as Simpson's paradox. Proposals for causality-based definitions of fairness can correctly model some of these situations, but they rely on background knowledge of the underlying causal models. In this paper, we formalize the situation as a database repair problem, proving sufficient conditions for fair classifiers in terms of admissible variables as opposed to a complete causal model. We show that these conditions correctly capture subtle fairness violations. We then use these conditions as the basis for database repair algorithms that provide provable fairness guarantees about classifiers trained on their training labels. We demonstrate the effectiveness of our proposed techniques with experimental results.
DOI: 10.1145/3318464.3380573
发表时间: 2019-11
期刊: Proceedings of the 2020 ACM SIGMOD International Conference on Management of Data
影响因子: --
作者:
Batya Kenig;Pranay Mundra;G. Prasad;Babak Salimi;Dan Suciu
通讯作者: Batya Kenig;Pranay Mundra;G. Prasad;Babak Salimi;Dan Suciu
DOI: 10.1145/3318464.3389759
发表时间: 2020-04
期刊: Proceedings of the 2020 ACM SIGMOD International Conference on Management of Data
影响因子: --
作者:
Babak Salimi;Harsh Parikh;Moe Kayali;Sudeepa Roy;L. Getoor;Dan Suciu
通讯作者: Babak Salimi;Harsh Parikh;Moe Kayali;Sudeepa Roy;L. Getoor;Dan Suciu
大数据警务的不同影响
DOI: --
发表时间: 2017
期刊: Georgia law review
影响因子: --
作者:
Selbst, Andrew D.
通讯作者: Selbst, Andrew D.