论文标题
快速公平回归通过有效的共同信息近似
Fast Fair Regression via Efficient Approximations of Mutual Information
论文作者
论文摘要
迄今为止，大多数算法公平性的工作都集中在离散的结果上，例如决定是否授予某人贷款。在这些分类设置中，可以通过比较亚种群之间的结果率来直接衡量群体公平标准，例如独立性，分离和充分性。但是，许多重要的问题需要预测实现的结果，例如风险评分或保险费。在这种回归设置中，测量组公平标准在计算上具有挑战性，因为它需要估计条件概率密度函数之间的信息理论差异。本文介绍了从其（条件）相互信息定义的回归模型的独立性，分离和足够群体公平标准的快速近似，并使用近似临近人在正规风险最小化框架内执行公平性。现实世界数据集中的实验表明，尽管具有出色的计算效率，但我们的算法仍显示最先进的准确性/公平性权衡。
Most work in algorithmic fairness to date has focused on discrete outcomes, such as deciding whether to grant someone a loan or not. In these classification settings, group fairness criteria such as independence, separation and sufficiency can be measured directly by comparing rates of outcomes between subpopulations. Many important problems however require the prediction of a real-valued outcome, such as a risk score or insurance premium. In such regression settings, measuring group fairness criteria is computationally challenging, as it requires estimating information-theoretic divergences between conditional probability density functions. This paper introduces fast approximations of the independence, separation and sufficiency group fairness criteria for regression models from their (conditional) mutual information definitions, and uses such approximations as regularisers to enforce fairness within a regularised risk minimisation framework. Experiments in real-world datasets indicate that in spite of its superior computational efficiency our algorithm still displays state-of-the-art accuracy/fairness tradeoffs.
