On preserving non-discrimination when combining expert advice
On preserving non-discrimination when combining expert advice
复制标题
关于在结合专家建议时保持非歧视
DOI：
--
复制
发表时间：
2018
期刊：
影响因子：
--
通讯作者：
N. Srebro
中科院分区：
文献类型：
--
作者：
Avrim Blum;Suriya Gunasekar;Thodoris Lykouris;N. Srebro
我们研究了顺序决策和避免歧视受保护群体之间的相互作用，当例子到达网上，不遵循分布假设。我们考虑经典在线学习的最基本扩展：“给定一类预测器，它们在特定度量方面是单独非歧视性的，我们如何将它们联合收割机组合起来，在保持非歧视性的同时，表现得与最佳预测器一样好？“令人惊讶的是，我们表明，这项任务是无法实现的流行概念的“均衡的赔率”，需要平等的假阴性率和平等的假阳性率跨组。在积极的一面，对于另一个概念的非歧视，“均衡的错误率”，我们表明，运行单独的实例的经典乘法权重算法为每个组实现这一保证。有趣的是，即使对于这个概念，我们表明，具有更强的性能保证比乘法权重的算法不能保持非歧视性。
We study the interplay between sequential decision making and avoiding discrimination against protected groups, when examples arrive online and do not follow distributional assumptions. We consider the most basic extension of classical online learning: "Given a class of predictors that are individually non-discriminatory with respect to a particular metric, how can we combine them to perform as well as the best predictor, while preserving non-discrimination?" Surprisingly we show that this task is unachievable for the prevalent notion of "equalized odds" that requires equal false negative rates and equal false positive rates across groups. On the positive side, for another notion of non-discrimination, "equalized error rates", we show that running separate instances of the classical multiplicative weights algorithm for each group achieves this guarantee. Interestingly, even for this notion, we show that algorithms with stronger performance guarantees than multiplicative weights cannot preserve non-discrimination.