We address the problem of uncertainty calibration and introduce a novel calibration method, Parametrized Temperature Scaling (PTS). Standard deep neural networks typically yield uncalibrated predictions, which can be transformed into calibrated confidence scores using post-hoc calibration methods. In this contribution, we demonstrate that the performance of accuracy-preserving state-of-the-art post-hoc calibrators is limited by their intrinsic expressive power. We generalize temperature scaling by computing prediction-specific temperatures, parameterized by a neural network. We show with extensive experiments that our novel accuracy-preserving approach consistently outperforms existing algorithms across a large number of model architectures, datasets and metrics.
@article{arxiv.2102.12182,
title = {Parameterized Temperature Scaling for Boosting the Expressive Power in Post-Hoc Uncertainty Calibration},
author = {Christian Tomani and Daniel Cremers and Florian Buettner},
journal= {arXiv preprint arXiv:2102.12182},
year = {2022}
}
Comments
In Proceedings of the European Conference on Computer Vision (ECCV), 2022. Code available at https://github.com/tochris/pts-uncertainty