Domain Feature Collapse: Implications for Out-of-Distribution Detection and Solutions

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Yang, Hong, Kar, Devroop, Yu, Qi, Ororbia, Alex, Desell, Travis
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866912962121302016
author Yang, Hong
Kar, Devroop
Yu, Qi
Ororbia, Alex
Desell, Travis
author_facet Yang, Hong
Kar, Devroop
Yu, Qi
Ororbia, Alex
Desell, Travis
contents Why do state-of-the-art OOD detection methods exhibit catastrophic failure when models are trained on single-domain datasets? We provide the first theoretical explanation for this phenomenon through the lens of information theory. We prove that supervised learning on single-domain data inevitably produces domain feature collapse -- representations where I(x_d; z) = 0, meaning domain-specific information is completely discarded. This is a fundamental consequence of information bottleneck optimization: models trained on single domains (e.g., medical images) learn to rely solely on class-specific features while discarding domain features, leading to catastrophic failure when detecting out-of-domain samples (e.g., achieving only 53% FPR@95 on MNIST). We extend our analysis using Fano's inequality to quantify partial collapse in practical scenarios. To validate our theory, we introduce Domain Bench, a benchmark of single-domain datasets, and demonstrate that preserving I(x_d; z) > 0 through domain filtering (using pretrained representations) resolves the failure mode. While domain filtering itself is conceptually straightforward, its effectiveness provides strong empirical evidence for our information-theoretic framework. Our work explains a puzzling empirical phenomenon, reveals fundamental limitations of supervised learning in narrow domains, and has broader implications for transfer learning and when to fine-tune versus freeze pretrained models.
format Preprint
id arxiv_https___arxiv_org_abs_2512_04034
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Domain Feature Collapse: Implications for Out-of-Distribution Detection and Solutions
Yang, Hong
Kar, Devroop
Yu, Qi
Ororbia, Alex
Desell, Travis
Machine Learning
Why do state-of-the-art OOD detection methods exhibit catastrophic failure when models are trained on single-domain datasets? We provide the first theoretical explanation for this phenomenon through the lens of information theory. We prove that supervised learning on single-domain data inevitably produces domain feature collapse -- representations where I(x_d; z) = 0, meaning domain-specific information is completely discarded. This is a fundamental consequence of information bottleneck optimization: models trained on single domains (e.g., medical images) learn to rely solely on class-specific features while discarding domain features, leading to catastrophic failure when detecting out-of-domain samples (e.g., achieving only 53% FPR@95 on MNIST). We extend our analysis using Fano's inequality to quantify partial collapse in practical scenarios. To validate our theory, we introduce Domain Bench, a benchmark of single-domain datasets, and demonstrate that preserving I(x_d; z) > 0 through domain filtering (using pretrained representations) resolves the failure mode. While domain filtering itself is conceptually straightforward, its effectiveness provides strong empirical evidence for our information-theoretic framework. Our work explains a puzzling empirical phenomenon, reveals fundamental limitations of supervised learning in narrow domains, and has broader implications for transfer learning and when to fine-tune versus freeze pretrained models.
title Domain Feature Collapse: Implications for Out-of-Distribution Detection and Solutions
topic Machine Learning
url https://arxiv.org/abs/2512.04034