An Elementary Predictor Obtaining $2\sqrt{T}+1$ Distance to Calibration
Machine Learning
2024-10-08 v2 Data Structures and Algorithms
Machine Learning
Abstract
Blasiok et al. [2023] proposed distance to calibration as a natural measure of calibration error that unlike expected calibration error (ECE) is continuous. Recently, Qiao and Zheng [2024] gave a non-constructive argument establishing the existence of an online predictor that can obtain distance to calibration in the adversarial setting, which is known to be impossible for ECE. They leave as an open problem finding an explicit, efficient algorithm. We resolve this problem and give an extremely simple, efficient, deterministic algorithm that obtains distance to calibration error at most .
Keywords
Cite
@article{arxiv.2402.11410,
title = {An Elementary Predictor Obtaining $2\sqrt{T}+1$ Distance to Calibration},
author = {Eshwar Ram Arunachaleswaran and Natalie Collina and Aaron Roth and Mirah Shi},
journal= {arXiv preprint arXiv:2402.11410},
year = {2024}
}