On the Fairness of Disentangled Representations
On the Fairness of Disentangled Representations
复制标题
DOI：
--
复制
发表时间：
2019-05
期刊：
影响因子：
5.2
通讯作者：
Francesco Locatello;G. Abbati;Tom Rainforth;Stefan Bauer;B. Scholkopf;Olivier Bachem
中科院分区：
文献类型：
--
作者：
Francesco Locatello;G. Abbati;Tom Rainforth;Stefan Bauer;B. Scholkopf;Olivier Bachem
最近，人们对学习解纠缠表示产生了浓厚的兴趣，因为它们有望提高可解释性，推广到看不见的场景，并更快地学习下游任务。在本文中，我们调查的有用性不同的概念，以提高下游预测任务的公平性的基础上表示的解纠缠。我们考虑这样一种设置，其目标是根据高维观测（如图像）的学习表示来预测目标变量，这些高维观测依赖于目标变量和\n {未观测}敏感变量。我们表明，在这种情况下，最佳和经验的预测可能是不公平的，即使目标变量和敏感变量是独立的。通过分析超过\num{12600}个经过训练的最先进的解纠缠模型的表示，我们观察到几个解纠缠分数始终与增加的公平性相关，这表明当没有观察到敏感变量时，解纠缠可能是鼓励公平性的有用属性。
Recently there has been a significant interest in learning disentangled representations, as they promise increased interpretability, generalization to unseen scenarios and faster learning on downstream tasks. In this paper, we investigate the usefulness of different notions of disentanglement for improving the fairness of downstream prediction tasks based on representations. We consider the setting where the goal is to predict a target variable based on the learned representation of high-dimensional observations (such as images) that depend on both the target variable and an \emph{unobserved} sensitive variable. We show that in this setting both the optimal and empirical predictions can be unfair, even if the target variable and the sensitive variable are independent. Analyzing the representations of more than \num{12600} trained state-of-the-art disentangled models, we observe that several disentanglement scores are consistently correlated with increased fairness, suggesting that disentanglement may be a useful property to encourage fairness when sensitive variables are not observed.