Adversarial Training with Rectified Rejection
Lists
Only you see this
Abstract
Adversarial training (AT) is one of the most effective strategies for promoting model robustness, whereas even the state-of-the-art adversarially trained models struggle to exceed 65% robust test accuracy on CIFAR-10 without additional data, which is far from practical. A natural way to improve beyond this accuracy bottleneck is to introduce a rejection option, where confidence is a commonly used certainty proxy. However, the vanilla confidence can overestimate the model certainty if the input is wrongly classified. To this end, we propose to use true confidence (T-Con) (i.e., predicted probability of the true class) as a certainty oracle, and learn to predict T-Con by rectifying confidence. Intriguingly, we prove that under mild conditions, a rectified confidence (R-Con) rejector and a confidence rejector can be coupled to distinguish any wrongly classified input from correctly classified ones. We also quantify that training R-Con to be aligned with T-Con could be an easier task than learning robust classifiers. In our experiments, we evaluate our rectified rejection (RR) module on CIFAR-10, CIFAR-10-C, and CIFAR-100 under several attacks, and demonstrate that the RR module is well compatible with different AT frameworks on improving robustness, with little extra computation.
BibTeX
@inproceedings{pang2021adversarial,
author = {Tianyu Pang and Huishuai Zhang and Di He and Yinpeng Dong and Hang Su and Wei Chen and Jun Zhu and Tie-Yan Liu},
title = {{Adversarial Training with Rectified Rejection}},
booktitle = {IEEE/CVF Conference on Computer Vision and Pattern Recognition},
year = {2021},
}| Stratified Adversarial Robustness with Rejection | arXiv · 2023 |
| Two Heads are Actually Better than One: Towards Better Adversarial Robustness via Transduction and Rejection | arXiv · 2023 |
| Deep neural rejection against adversarial examples | Eurasip J. Inf. Secur · 2020 |
| Confidence-Calibrated Adversarial Training: Generalizing to Unseen Attacks | arXiv · 2019 |
| Robustness to Adversarial Examples through an Ensemble of Specialists | Workshops @ ICLR · 2017 |
| Robust Models are less Over-Confident | arXiv · 2022 |