Saved in:
Bibliographic Details
Main Author: Salem, Mohamed
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2603.06609
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917319148568576
author Salem, Mohamed
author_facet Salem, Mohamed
contents Modern machine learning models are highly expressive but notoriously difficult to analyze statistically. In particular, while black-box predictors can achieve strong empirical performance, they rarely provide valid hypothesis tests or p-values for assessing whether individual features contain information about a target variable. This article presents a practical approach to feature-level hypothesis testing that combines the Conditional Randomization Test (CRT) with TabPFN, a probabilistic foundation model for tabular data. The resulting procedure yields finite-sample valid p-values for conditional feature relevance, even in nonlinear and correlated settings, without requiring model retraining or parametric assumptions.
format Preprint
id arxiv_https___arxiv_org_abs_2603_06609
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Valid Feature-Level Inference for Tabular Foundation Models via the Conditional Randomization Test
Salem, Mohamed
Machine Learning
Modern machine learning models are highly expressive but notoriously difficult to analyze statistically. In particular, while black-box predictors can achieve strong empirical performance, they rarely provide valid hypothesis tests or p-values for assessing whether individual features contain information about a target variable. This article presents a practical approach to feature-level hypothesis testing that combines the Conditional Randomization Test (CRT) with TabPFN, a probabilistic foundation model for tabular data. The resulting procedure yields finite-sample valid p-values for conditional feature relevance, even in nonlinear and correlated settings, without requiring model retraining or parametric assumptions.
title Valid Feature-Level Inference for Tabular Foundation Models via the Conditional Randomization Test
topic Machine Learning
url https://arxiv.org/abs/2603.06609