Is There a Trade-Off Between Fairness and Accuracy? A Perspective Using Mismatched Hypothesis Testing
Is There a Trade-Off Between Fairness and Accuracy? A Perspective Using Mismatched Hypothesis Testing
复制标题
公平性和准确性之间是否需要权衡？
DOI：
--
复制
发表时间：
2020
期刊：
影响因子：
--
通讯作者：
Kush R. Varshney
中科院分区：
文献类型：
--
作者：
Sanghamitra Dutta;Dennis Wei;Hazar Yueksel;Pin;Sijia Liu;Kush R. Varshney
在关于机器学习公平性的现有文献中，准确性和公平性之间的权衡几乎被认为是给定的。然而，并不是说准确性会随着公平性的增加而降低。这项工作的新颖之处在于，我们通过错配假设检验的视角来检查公平分类：当给定两个有偏差的错配分布时，试图找到一个分类器来区分两个理想分布。利用信息理论中的工具Chernoff信息，我们从理论上证明，与普遍的看法相反，总是存在理想分布，使最佳公平性和准确性（相对于理想分布）同时实现：没有权衡。此外，相同的分类筛选器在相对于理想分布而言缺乏权衡，而在相对于给定（可能有偏差）数据集测量精度时产生权衡。为了补充我们的主要结果，我们制定了一个优化来找到理想分布，并推导出基本限制来解释为什么在给定的有偏差数据集上存在权衡。我们还推导了在哪些条件下主动数据收集可以缓解现实世界中的公平-准确性权衡。我们的结果使我们认为，衡量反映偏差的数据的准确性是有问题的，相反，我们应该考虑相对于理想的、无偏的数据的准确性。
A trade-off between accuracy and fairness is almost taken as a given in the existing literature on fairness in machine learning. Yet, it is not preordained that accuracy should decrease with increased fairness. Novel to this work, we examine fair classification through the lens of mis-matched hypothesis testing : trying to find a clas-sifier that distinguishes between two ideal distributions when given two mismatched distributions that are biased. Using Chernoff information, a tool in information theory, we theoretically demonstrate that, contrary to popular belief, there always exist ideal distributions such that optimal fairness and accuracy (with respect to the ideal distributions) are achieved simultaneously: there is no trade-off. Moreover, the same clas-sifier yields the lack of a trade-off with respect to ideal distributions while yielding a trade-off when accuracy is measured with respect to the given (possibly biased) dataset. To complement our main result, we formulate an optimization to find ideal distributions and derive fundamental limits to explain why a trade-off exists on the given biased dataset. We also derive conditions under which active data collection can alleviate the fairness-accuracy trade-off in the real world. Our results lead us to contend that it is problematic to measure accuracy with respect to data that reflects bias, and instead, we should be considering accuracy with respect to ideal, unbiased data.
DOI：
--
发表时间：
2019
期刊：
EC
影响因子：
--
作者：
Garg, Sumegha;Kim, Michael P.;Reingold, Omer
通讯作者：
Reingold, Omer