Research map / Learning theory and algorithms / Fairness

An Efficient Geometric Characterization of Robust Fair Learning

NeurIPS 2026 · Learning theory and algorithms › Fairness · approach: Fairness-constrained classification

Read the paperCheck your idea against 108,967 papers

AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Studies whether the optimal fair classifier from a hypothesis class remains accurate under malicious distribution shifts, and gives an efficient geometric characterization of robust fair learning.

Problem

Optimal classifiers under exact fairness constraints may be non-robust: their accuracy can change drastically under small distribution shifts. The paper asks whether the optimal fair classifier from a hypothesis class is robust to malicious shifts and how to efficiently characterize and compute robu

Approach

A geometric characterization of robust fair learning that yields an efficient way to identify robust fair classifiers from a given hypothesis class, in contrast to prior work showing exact-fairness optima can be fragile.

Surprising findings

• The optimal classifier under exact fairness constraints from a hypothesis class may not be robust; its accuracy can change drastically under small distribution shifts.

Closest papers by meaning

• Optimal Fair Learning Robust to Adversarial Distribution Shift (ICML 2025 · similarity 0.92)
• Ensuring Fairness Beyond the Training Data (NeurIPS 2020 · similarity 0.91)
• Enhancing Robust Fairness via Confusional Spectral Regularization (ICLR 2025 · similarity 0.87)
• Efficient Fairness-Performance Pareto Front Computation (NeurIPS 2025 · similarity 0.87)
• How Far Can Fairness Constraints Help Recover From Biased Data? (ICML 2024 · similarity 0.86)
• Geometry of Relaxed Fair Regression: A Unified Framework for Aware and Unaware Settings (NeurIPS 2026 · similarity 0.86)