Saved in:
Bibliographic Details
Main Authors: Bayle, Alexandre, Janson, Lucas, Mackey, Lester
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2508.04409
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918326369779712
author Bayle, Alexandre
Janson, Lucas
Mackey, Lester
author_facet Bayle, Alexandre
Janson, Lucas
Mackey, Lester
contents Cross-validation (CV) is known to provide asymptotically exact tests and confidence intervals for model improvement but only when the model comparison is relatively stable. Surprisingly, we prove that even simple, individually stable models can generate relatively unstable comparisons, calling into question the validity of CV inference. Specifically, we show that the Lasso and its close cousin, soft-thresholding, generate relatively unstable comparisons and invalid CV inferences, even in the most favorable of learning settings and when both models are individually stable. These findings highlight the importance of verifying relative stability before deploying CV for model comparison.
format Preprint
id arxiv_https___arxiv_org_abs_2508_04409
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle The Relative Instability of Model Comparison with Cross-validation
Bayle, Alexandre
Janson, Lucas
Mackey, Lester
Machine Learning
Cross-validation (CV) is known to provide asymptotically exact tests and confidence intervals for model improvement but only when the model comparison is relatively stable. Surprisingly, we prove that even simple, individually stable models can generate relatively unstable comparisons, calling into question the validity of CV inference. Specifically, we show that the Lasso and its close cousin, soft-thresholding, generate relatively unstable comparisons and invalid CV inferences, even in the most favorable of learning settings and when both models are individually stable. These findings highlight the importance of verifying relative stability before deploying CV for model comparison.
title The Relative Instability of Model Comparison with Cross-validation
topic Machine Learning
url https://arxiv.org/abs/2508.04409