Saved in:
Bibliographic Details
Main Author: Mehta, Sushant
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2511.07485
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908641558265856
author Mehta, Sushant
author_facet Mehta, Sushant
contents Machine learning systems exhibit diverse failure modes: unfairness toward protected groups, brittleness to spurious correlations, poor performance on minority sub-populations, which are typically studied in isolation by distinct research communities. We propose a unifying theoretical framework that characterizes when different bias mechanisms produce quantitatively equivalent effects on model performance. By formalizing biases as violations of conditional independence through information-theoretic measures, we prove formal equivalence conditions relating spurious correlations, subpopulation shift, class imbalance, and fairness violations. Our theory predicts that a spurious correlation of strength $α$ produces equivalent worst-group accuracy degradation as a sub-population imbalance ratio $r \approx (1+α)/(1-α)$ under feature overlap assumptions. Empirical validation in six datasets and three architectures confirms that predicted equivalences hold within the accuracy of the worst group 3\%, enabling the principled transfer of debiasing methods across problem domains. This work bridges the literature on fairness, robustness, and distribution shifts under a common perspective.
format Preprint
id arxiv_https___arxiv_org_abs_2511_07485
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle When Are Learning Biases Equivalent? A Unifying Framework for Fairness, Robustness, and Distribution Shift
Mehta, Sushant
Machine Learning
Artificial Intelligence
Machine learning systems exhibit diverse failure modes: unfairness toward protected groups, brittleness to spurious correlations, poor performance on minority sub-populations, which are typically studied in isolation by distinct research communities. We propose a unifying theoretical framework that characterizes when different bias mechanisms produce quantitatively equivalent effects on model performance. By formalizing biases as violations of conditional independence through information-theoretic measures, we prove formal equivalence conditions relating spurious correlations, subpopulation shift, class imbalance, and fairness violations. Our theory predicts that a spurious correlation of strength $α$ produces equivalent worst-group accuracy degradation as a sub-population imbalance ratio $r \approx (1+α)/(1-α)$ under feature overlap assumptions. Empirical validation in six datasets and three architectures confirms that predicted equivalences hold within the accuracy of the worst group 3\%, enabling the principled transfer of debiasing methods across problem domains. This work bridges the literature on fairness, robustness, and distribution shifts under a common perspective.
title When Are Learning Biases Equivalent? A Unifying Framework for Fairness, Robustness, and Distribution Shift
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2511.07485