We investigate fairness in classification, where automated decisions are made for individuals from different protected groups. In high-consequence scenarios, decision errors can disproportionately affect certain protected groups, leading to unfair outcomes. To address this issue, we propose a fairness-adjusted selective inference (FASI) framework and develop data-driven algorithms that achieve statistical parity by controlling and equalizing the false selection rate (FSR) among protected groups. Our FASI algorithm operates by converting the outputs of black-box classifiers into R-values, which are both intuitive and computationally efficient. The selection rules based on R-values, which effectively mitigate disparate impacts on protected groups, are provably valid for FSR control in finite samples. We demonstrate the numerical performance of our approach through both simulated and real data.

翻译：我们研究了分类中的公平性问题，其中自动化决策针对来自不同受保护群体的个体做出。在高风险场景中，决策错误可能对某些受保护群体造成不成比例的影响，从而导致不公平的结果。为了解决这一问题，我们提出了一种公平性调整的选择性推断（FASI）框架，并开发了基于数据的算法，通过控制并均衡受保护群体间的错误选择率（FSR）来实现统计均等。我们的FASI算法通过将黑箱分类器的输出转换为R值来运作，这种方法既直观又计算高效。基于R值的选择规则能有效减轻对受保护群体的差异化影响，并且在有限样本下可证明有效控制FSR。我们通过模拟数据和真实数据展示了该方法的数值性能。