Saved in:
Bibliographic Details
Main Authors: Glaser, Pierre, Widmann, David, Lindsten, Fredrik, Gretton, Arthur
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2510.14711
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912652774604800
author Glaser, Pierre
Widmann, David
Lindsten, Fredrik
Gretton, Arthur
author_facet Glaser, Pierre
Widmann, David
Lindsten, Fredrik
Gretton, Arthur
contents We introduce the Kernel Calibration Conditional Stein Discrepancy test (KCCSD test), a non-parametric, kernel-based test for assessing the calibration of probabilistic models with well-defined scores. In contrast to previous methods, our test avoids the need for possibly expensive expectation approximations while providing control over its type-I error. We achieve these improvements by using a new family of kernels for score-based probabilities that can be estimated without probability density samples, and by using a conditional goodness-of-fit criterion for the KCCSD test's U-statistic. We demonstrate the properties of our test on various synthetic settings.
format Preprint
id arxiv_https___arxiv_org_abs_2510_14711
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Fast and Scalable Score-Based Kernel Calibration Tests
Glaser, Pierre
Widmann, David
Lindsten, Fredrik
Gretton, Arthur
Machine Learning
We introduce the Kernel Calibration Conditional Stein Discrepancy test (KCCSD test), a non-parametric, kernel-based test for assessing the calibration of probabilistic models with well-defined scores. In contrast to previous methods, our test avoids the need for possibly expensive expectation approximations while providing control over its type-I error. We achieve these improvements by using a new family of kernels for score-based probabilities that can be estimated without probability density samples, and by using a conditional goodness-of-fit criterion for the KCCSD test's U-statistic. We demonstrate the properties of our test on various synthetic settings.
title Fast and Scalable Score-Based Kernel Calibration Tests
topic Machine Learning
url https://arxiv.org/abs/2510.14711