Trust, but Verify: Using Self-Supervised Probing to Improve Trustworthiness
Abstract
Trustworthy machine learning is of primary importance to the practical deployment of deep learning models. While state-of-the-art models achieve astonishingly good performance in terms of accuracy, recent literature reveals that their predictive confidence scores unfortunately cannot be trusted: e.g., they are often overconfident when wrong predictions are made, or so even for obvious outliers. In this paper, we introduce a new approach of self-supervised probing, which enables us to check and mitigate the overconfidence issue for a trained model, thereby improving its trustworthiness. We provide a simple yet effective framework, which can be flexibly applied to existing trustworthiness-related methods in a plug-and-play manner. Extensive experiments on three trustworthiness-related tasks (misclassification detection, calibration and out-of-distribution detection) across various benchmarks verify the effectiveness of our proposed probing framework.
Cite
@article{arxiv.2302.02628,
title = {Trust, but Verify: Using Self-Supervised Probing to Improve Trustworthiness},
author = {Ailin Deng and Shen Li and Miao Xiong and Zhirui Chen and Bryan Hooi},
journal= {arXiv preprint arXiv:2302.02628},
year = {2023}
}
Comments
European Conference on Computer Vision 2022