Fairness Without Demographics in Repeated Loss Minimization
Tatsunori B. Hashimoto Megha Srivastava Hongseok Namkoong Percy Liang
Abstract
Machine learning models (e.g., speech recognizers) are usually trained to minimize average loss, which results in representation disparity—minority groups (e.g., non-native speakers) contribute less to the training objective and thus tend to suffer higher loss. Worse, as model accuracy affects user retention, a minority group can shrink over time. In this paper, we first show that the status quo of empirical risk minimization (ERM) amplifies representation disparity over time, which can even make initially fair models unfair. To mitigate this, we develop an approach based on distributionally robust optimization (DRO), which minimizes the worst case risk over all distributions close to the empirical distribution. We prove that this approach controls the risk of the minority group at each time step, in the spirit of Rawlsian distributive justice, while remaining oblivious to the identity of the groups. We demonstrate that DRO prevents disparity amplification on examples where ERM fails, and show improvements in minority group user satisfaction in a real-world text autocomplete task.
中文速览
主流机器学习模型通常以最小化平均损失为目标来训练，这会导致少数群体（如带口音的语音识别用户）的错误率系统性偏高——即"表示差异（representation disparity）"问题，而更糟糕的是，这些群体因体验差而流失用户、数据越来越少，下一轮重训后模型对他们反而更差，形成恶性循环。论文首先从理论上证明了经验风险最小化（ERM）会随时间不断放大这种差异，甚至把初始公平的模型变得不公平。为此，作者提出用分布鲁棒优化（DRO）来替代 ERM——在靠近经验分布的一个卡方散度球内最小化最坏情况下的风险，且整个过程不需要知道用户的群体身份信息。理论分析证明该方法能在每个时间步控制少数群体的风险上界，实验和基于 Amazon Mechanical Turk 的真实文本自动补全任务也验证了 DRO 相比 ERM 能有效遏制差异放大、提升少数群体满意度，对于公平地部署和持续迭代机器学习系统具有重要意义。
原文 arXiv:1806.08010;中英对照 + 大白话阅读 https://aha.fim.ai/paper/1806.08010v2