论文标题
通过速率最大化学习公平表示
Learning Fair Representations via Rate-Distortion Maximization
论文作者
论文摘要
通过机器学习模型学到的文本表示通常编码用户的不良人口统计信息。基于这些表示形式的预测模型可以依靠此类信息，从而产生偏见的决策。我们提出了一种新颖的偏见技术，即公平意识的速率最大化（FARM），该技术使用速率依赖函数来消除受保护的信息，以表示属于相同受保护的属性类别的实例。 Farm能够在有或没有目标任务的情况下进行DEBIAS表示。也可以对农场同时删除有关多个受保护属性的信息。经验评估表明，Farm在几个数据集上实现了最新的性能，并且学会的表示形式泄漏了受保护的属性信息明显减少，以防止非线性探测网络攻击。
Text representations learned by machine learning models often encode undesirable demographic information of the user. Predictive models based on these representations can rely on such information, resulting in biased decisions. We present a novel debiasing technique, Fairness-aware Rate Maximization (FaRM), that removes protected information by making representations of instances belonging to the same protected attribute class uncorrelated, using the rate-distortion function. FaRM is able to debias representations with or without a target task at hand. FaRM can also be adapted to remove information about multiple protected attributes simultaneously. Empirical evaluations show that FaRM achieves state-of-the-art performance on several datasets, and learned representations leak significantly less protected attribute information against an attack by a non-linear probing network.
