arXivDaily arXiv每日学术速递 周一至周五更新
北美退伍军人节(Veterans Day),12 日无更新
arXiv 2005.11720cs.LGmath.STstat.MLstat.TH
收藏

统计学习中的公平性投影

Projection to Fairness in Statistical Learning

Massachusetts Institute of Technology(麻省理工学院);University Toulouse 3(图卢兹第三大学)

Thibaut Le Gouic, Jean-Michel Loubes, Philippe Rigollet

摘要:

在回归的背景下,我们考虑的基本问题是使估计公平,同时尽可能保持其预测精度。为此,我们将其对公平性的投影定义为反映预测准确性的最接近公平估计量。我们的方法利用最优运输工具来有效地构建任何给定估计量的公平性投影,作为一个简单的后处理步骤。此外,我们的方法精确地量化了公平性的成本,以预测准确性衡量。

英文摘要:

In the context of regression, we consider the fundamental question of making an estimator fair while preserving its prediction accuracy as much as possible. To that end, we define its projection to fairness as its closest fair estimator in a sense that reflects prediction accuracy. Our methodology leverages tools from optimal transport to construct efficiently the projection to fairness of any given estimator as a simple post-processing step. Moreover, our approach precisely quantifies the cost of fairness, measured in terms of prediction accuracy.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑