[Submitted on 29 Sep 2026 (v1), last revised 30 Oct 2026 (this version, v2)]
Abstract:We view fairness as a property of distributional stability. Rather than assessing a predictor under a fixed data distribution, we study how its predictions change under perturbations that modify the composition of protected groups. A predictor is fair if it remains stable under such shifts. Under this perspective, several classical notions of fairness arise as stability with respect to specific perturbations, with the associated unfairness gap given by a Lipschitz constant of a prediction-rate functional. This formulation also yields guarantees that hold uniformly over a range of demographic compositions at test time, without requiring knowledge of the deployment distribution. It leads to a learning procedure based on convex combinations of reweighted predictors, formulated as a second-order cone program, for which we establish generalization bounds. Experiments on standard benchmarks illustrate the approach.
Submission history
From: Charlotte Laclau [view email]
[v1] Tue, 29 Sep 2026 09:00:49 UTC (361 KB)
[v2] Fri, 30 Oct 2026 10:42:03 UTC (545 KB)
