[论文解读] Fair Classification and Social Welfare

Lily Hu, Yiling Chen|arXiv (Cornell University)|May 1, 2019

Ethics and Social Impacts of AI参考文献 34被引用 4

一句话总结

本文将机器学习分类问题重新构架为社会福利最大化问题，将公平性约束与效用结果相联系。研究发现，更严格的公平性约束（如基于平等的公平性）可能降低个人与群体的整体福利，挑战了‘更公平’总是对弱势群体更有利的假设。

ABSTRACT

Now that machine learning algorithms lie at the center of many resource allocation pipelines, computer scientists have been unwittingly cast as partial social planners. Given this state of affairs, important questions follow. What is the relationship between fairness as defined by computer scientists and notions of social welfare? In this paper, we present a welfare-based analysis of classification and fairness regimes. We translate a loss minimization program into a social welfare maximization problem with a set of implied welfare weights on individuals and groups--weights that can be analyzed from a distribution justice lens. In the converse direction, we ask what the space of possible labelings is for a given dataset and hypothesis class. We provide an algorithm that answers this question with respect to linear hyperplanes in $\mathbb{R}^d$ that runs in $O(n^dd)$. Our main findings on the relationship between fairness criteria and welfare center on sensitivity analyses of fairness-constrained empirical risk minimization programs. We characterize the ranges of $Δε$ perturbations to a fairness parameter $ε$ that yield better, worse, and neutral outcomes in utility for individuals and by extension, groups. We show that applying more strict fairness criteria that are codified as parity constraints, can worsen welfare outcomes for both groups. More generally, always preferring "more fair" classifiers does not abide by the Pareto Principle---a fundamental axiom of social choice theory and welfare economics. Recent work in machine learning has rallied around these notions of fairness as critical to ensuring that algorithmic systems do not have disparate negative impact on disadvantaged social groups. By showing that these constraints often fail to translate into improved outcomes for these groups, we cast doubt on their effectiveness as a means to ensure justice.

研究动机与目标

• 分析机器学习中的公平性与福利经济学之间的关系。
• 研究当前形式的公平性约束是否真的能改善弱势群体的福利结果。
• 构建一个将损失最小化转化为社会福利最大化的框架，并引入可解释的福利权重。
• 刻画公平性约束扰动对分类器效用和群体结果的影响。
• 提出一种算法，用于计算线性超平面分类器下所有可实现的标记分配，从而实现对福利权衡的全面探索。

提出的方法

• 将经验风险最小化重新表述为社会计划者的优化问题，以最大化个体效用的加权和。
• 从最优分类结果中推导出福利权重，从而将分类器决策解释为社会资源的分配。
• 提出一种 O(ndd) 算法，用于计算在 d 维空间中线性超平面对所有可实现的标记分配。
• 应用参数规划与敏感性分析，研究公平性容忍度参数 ϵ 的扰动影响。
• 使用拉格朗日乘子与影子价格，衡量公平性约束对 ϵ 变化的局部敏感性。
• 通过扰动函数推导出福利损失的全局下界，从而实现对约束收紧或放松的非对称解释。

实验结果

研究问题

• RQ1 机器学习中的公平性约束如何影响个人与群体的社会福利结果？
• RQ2 更严格的公平性约束（如基于平等的公平性）是否可能导致群体福利结果恶化？
• RQ3 对公平性容忍度参数 ϵ 的扰动如何影响分类器效用与群体福利？
• RQ4 如何系统性地计算在给定假设类与数据集下所有可能的标记分配？
• RQ5 公平性度量在多大程度上与个体与群体福利的实际改善相一致？

主要发现

• 以平等约束形式编码的更严格公平性约束可能使个人与群体的福利结果恶化，这与‘更公平即更好’的假设相矛盾。
• 收紧公平性参数 ϵ（即施加更严格约束）会导致学习者损失增加，且无法保证弱势群体的福利结果得到改善。
• 当公平性约束的影子价格较高时，收紧 ϵ 会导致福利损失显著上升，表明存在较高的权衡代价。
• 当影子价格较低时，放松公平性约束仅导致损失小幅下降，表明对约束变化的敏感性具有非对称性。
• 在损失最小化下的最优分类器对应于一组特定隐含福利权重下的福利最大化分配，这些权重可从模型预测中推导得出。
• 存在 ∆ϵ < 0 的扰动，即收紧公平性约束，即使从非零 ϵ-公平解出发，也会导致分类器性能显著恶化。