Measuring Uncertainty Calibration
Abstract
We make two contributions to the problem of estimating the $L_1$ calibration error of a binary classifier from a finite dataset. First, we provide an upper bound for any classifier where the calibration function has bounded variation. Second, we provide a method of modifying any classifier so that its calibration error can be upper bounded efficiently without significantly impacting classifier performance and without any restrictive assumptions. All our results are non-asymptotic and distribution-free. We conclude by providing advice on how to measure calibration error in practice. Our methods yield practical procedures that can be run on real-world datasets with modest overhead.
Cite
Text
Ciosek et al. "Measuring Uncertainty Calibration." International Conference on Learning Representations, 2026.Markdown
[Ciosek et al. "Measuring Uncertainty Calibration." International Conference on Learning Representations, 2026.](https://mlanthology.org/iclr/2026/ciosek2026iclr-measuring/)BibTeX
@inproceedings{ciosek2026iclr-measuring,
title = {{Measuring Uncertainty Calibration}},
author = {Ciosek, Kamil and Felicioni, Nicolò and Ghiassian, Sina and Elenter, Juan and Tonolini, Francesco and Gustafsson, David and Garcia-Martin, Eva and Gonzalez, Carmen Barcena and Bertrand-Lalo, Raphaëlle},
booktitle = {International Conference on Learning Representations},
year = {2026},
url = {https://mlanthology.org/iclr/2026/ciosek2026iclr-measuring/}
}