Fast and Scalable Score-Based Kernel Calibration Tests

Abstract

We introduce the Kernel Calibration Conditional Stein Discrepancy test (KCCSD test), a nonparametric, 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.

Cite

Text

Glaser et al. "Fast and Scalable Score-Based Kernel Calibration Tests." Uncertainty in Artificial Intelligence, 2023.

Markdown

[Glaser et al. "Fast and Scalable Score-Based Kernel Calibration Tests." Uncertainty in Artificial Intelligence, 2023.](https://mlanthology.org/uai/2023/glaser2023uai-fast/)

BibTeX

@inproceedings{glaser2023uai-fast,
  title     = {{Fast and Scalable Score-Based Kernel Calibration Tests}},
  author    = {Glaser, Pierre and Widmann, David and Lindsten, Fredrik and Gretton, Arthur},
  booktitle = {Uncertainty in Artificial Intelligence},
  year      = {2023},
  pages     = {691-700},
  volume    = {216},
  url       = {https://mlanthology.org/uai/2023/glaser2023uai-fast/}
}