Doubly-Regressing Approach for Subgroup Fairness
Kyungseon Lee, Kunwoong Kim, Jihu Lee, Dongyoon Yang, Yongdai Kim
TL;DR
The paper tackles intersectional fairness when many sensitive attributes yield a combinatorial explosion of subgroups, causing data sparsity and heavy computation. It introduces subgroup-subset fairness and the supIPM metric to enforce distributional fairness on a selectively chosen set of active subgroups, including marginal groups, and develops the Doubly Regressing Adversarial Learning (DRAF) framework. A tractable surrogate based on $\textup{DR}^2$ enables a single discriminator to bound the supIPM across all chosen subgroup-subsets, with theoretical guarantees that the surrogate upper-bounds the fairness gap. Empirically, DRAF outperforms baselines, especially in highly sparse settings, while maintaining accuracy, and the work demonstrates a scalable path to robust, intersectional fairness in high-dimensional sensitive attribute settings.
Abstract
Algorithmic fairness is a socially crucial topic in real-world applications of AI. Among many notions of fairness, subgroup fairness is widely studied when multiple sensitive attributes (e.g., gender, race, age) are present. However, as the number of sensitive attributes grows, the number of subgroups increases accordingly, creating heavy computational burdens and data sparsity problem (subgroups with too small sizes). In this paper, we develop a novel learning algorithm for subgroup fairness which resolves these issues by focusing on subgroups with sufficient sample sizes as well as marginal fairness (fairness for each sensitive attribute). To this end, we formalize a notion of subgroup-subset fairness and introduce a corresponding distributional fairness measure called the supremum Integral Probability Metric (supIPM). Building on this formulation, we propose the Doubly Regressing Adversarial learning for subgroup Fairness (DRAF) algorithm, which reduces a surrogate fairness gap for supIPM with much less computation than directly reducing supIPM. Theoretically, we prove that the proposed surrogate fairness gap is an upper bound of supIPM. Empirically, we show that the DRAF algorithm outperforms baseline methods in benchmark datasets, specifically when the number of sensitive attributes is large so that many subgroups are very small.