Despite achieving enormous success in predictive accuracy for visual classification problems, deep neural networks (DNNs) suffer from providing overconfident probabilities on out-of-distribution (OOD) data. Yet, accurate uncertainty estimation is crucial for safe and reliable robot autonomy. In this paper, we evaluate popular calibration techniques for open-set conditions in a way that is distinctly different from the conventional evaluation of calibration methods on OOD data. Our results show that closed-set DNN calibration approaches are much less effective for open-set recognition, which highlights the need to develop new DNN calibration methods to address this problem.
@article{arxiv.2205.07160,
title = {Evaluating Uncertainty Calibration for Open-Set Recognition},
author = {Zongyao Lyu and Nolan B. Gutierrez and William J. Beksi},
journal= {arXiv preprint arXiv:2205.07160},
year = {2022}
}
Comments
To be presented at the 2022 IEEE International Conference on Robotics and Automation (ICRA) Workshop on Safe and Reliable Robot Autonomy under Uncertainty