Causality-Inspired Robustness for Nonlinear Models via Representation Learning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Šola, Marin, Bühlmann, Peter, Shen, Xinwei
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915292527984640
author Šola, Marin
Bühlmann, Peter
Shen, Xinwei
author_facet Šola, Marin
Bühlmann, Peter
Shen, Xinwei
contents Distributional robustness is a central goal of prediction algorithms due to the prevalent distribution shifts in real-world data. The prediction model aims to minimize the worst-case risk among a class of distributions, a.k.a., an uncertainty set. Causality provides a modeling framework with a rigorous robustness guarantee in the above sense, where the uncertainty set is data-driven rather than pre-specified as in traditional distributional robustness optimization. However, current causality-inspired robustness methods possess finite-radius robustness guarantees only in the linear settings, where the causal relationships among the covariates and the response are linear. In this work, we propose a nonlinear method under a causal framework by incorporating recent developments in identifiable representation learning and establish a distributional robustness guarantee. To our best knowledge, this is the first causality-inspired robustness method with such a finite-radius robustness guarantee in nonlinear settings. Empirical validation of the theoretical findings is conducted on both synthetic data and real-world single-cell data, also illustrating that finite-radius robustness is crucial.
format Preprint
id arxiv_https___arxiv_org_abs_2505_12868
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Causality-Inspired Robustness for Nonlinear Models via Representation Learning
Šola, Marin
Bühlmann, Peter
Shen, Xinwei
Machine Learning
Methodology
Distributional robustness is a central goal of prediction algorithms due to the prevalent distribution shifts in real-world data. The prediction model aims to minimize the worst-case risk among a class of distributions, a.k.a., an uncertainty set. Causality provides a modeling framework with a rigorous robustness guarantee in the above sense, where the uncertainty set is data-driven rather than pre-specified as in traditional distributional robustness optimization. However, current causality-inspired robustness methods possess finite-radius robustness guarantees only in the linear settings, where the causal relationships among the covariates and the response are linear. In this work, we propose a nonlinear method under a causal framework by incorporating recent developments in identifiable representation learning and establish a distributional robustness guarantee. To our best knowledge, this is the first causality-inspired robustness method with such a finite-radius robustness guarantee in nonlinear settings. Empirical validation of the theoretical findings is conducted on both synthetic data and real-world single-cell data, also illustrating that finite-radius robustness is crucial.
title Causality-Inspired Robustness for Nonlinear Models via Representation Learning
topic Machine Learning
Methodology
url https://arxiv.org/abs/2505.12868