When Shift Happens - Confounding Is to Blame

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Reddy, Abbavaram Gowtham, Rubio-Madrigal, Celia, Burkholz, Rebekka, Muandet, Krikamol
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915307740725248
author Reddy, Abbavaram Gowtham
Rubio-Madrigal, Celia
Burkholz, Rebekka
Muandet, Krikamol
author_facet Reddy, Abbavaram Gowtham
Rubio-Madrigal, Celia
Burkholz, Rebekka
Muandet, Krikamol
contents Distribution shifts introduce uncertainty that undermines the robustness and generalization capabilities of machine learning models. While conventional wisdom suggests that learning causal-invariant representations enhances robustness to such shifts, recent empirical studies present a counterintuitive finding: (i) empirical risk minimization (ERM) can rival or even outperform state-of-the-art out-of-distribution (OOD) generalization methods, and (ii) its OOD generalization performance improves when all available covariates, not just causal ones, are utilized. Drawing on both empirical and theoretical evidence, we attribute this phenomenon to hidden confounding. Shifts in hidden confounding induce changes in data distributions that violate assumptions commonly made by existing OOD generalization approaches. Under such conditions, we prove that effective generalization requires learning environment-specific relationships, rather than relying solely on invariant ones. Furthermore, we show that models augmented with proxies for hidden confounders can mitigate the challenges posed by hidden confounding shifts. These findings offer new theoretical insights and practical guidance for designing robust OOD generalization algorithms and principled covariate selection strategies.
format Preprint
id arxiv_https___arxiv_org_abs_2505_21422
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle When Shift Happens - Confounding Is to Blame
Reddy, Abbavaram Gowtham
Rubio-Madrigal, Celia
Burkholz, Rebekka
Muandet, Krikamol
Machine Learning
Distribution shifts introduce uncertainty that undermines the robustness and generalization capabilities of machine learning models. While conventional wisdom suggests that learning causal-invariant representations enhances robustness to such shifts, recent empirical studies present a counterintuitive finding: (i) empirical risk minimization (ERM) can rival or even outperform state-of-the-art out-of-distribution (OOD) generalization methods, and (ii) its OOD generalization performance improves when all available covariates, not just causal ones, are utilized. Drawing on both empirical and theoretical evidence, we attribute this phenomenon to hidden confounding. Shifts in hidden confounding induce changes in data distributions that violate assumptions commonly made by existing OOD generalization approaches. Under such conditions, we prove that effective generalization requires learning environment-specific relationships, rather than relying solely on invariant ones. Furthermore, we show that models augmented with proxies for hidden confounders can mitigate the challenges posed by hidden confounding shifts. These findings offer new theoretical insights and practical guidance for designing robust OOD generalization algorithms and principled covariate selection strategies.
title When Shift Happens - Confounding Is to Blame
topic Machine Learning
url https://arxiv.org/abs/2505.21422