Tao Li, Yingwen Wu, Sizhe Chen, Kun Fang and Xiaolin Huang

Paper: http://arxiv.org/abs/2111.12229

CVPR 2022 oral

Single-step adversarial training (AT) has received wide attention as it proved to be both efficient and robust. However, a serious problem of catastrophic overfitting exists, i.e., the robust accuracy against projected gradient descent (PGD) attack suddenly drops to 0% during the training. In this paper, we approach this problem from a novel perspective of optimization and firstly reveal the close link between the fast-growing gradient of each sample and overfitting, which can also be applied to understand robust overfitting in multi-step AT. To control the growth of the gradient, we propose a new AT method, Subspace Adversarial Training (Sub-AT), which constrains AT in a carefully extracted subspace. It successfully resolves both kinds of overfitting and significantly boosts the robustness. In subspace, we also allow single-step AT with larger steps and larger radius, further improving the robustness performance. As a result, we achieve state-of-the-art single-step AT performance. Without any regularization term, our single-step AT can reach over 51% robust accuracy against strong PGD-50 attack of radius 8/255 on CIFAR-10, reaching a competitive performance against standard multi-step PGD-10 AT with huge computational advantages.

Install required dependencies:

pip install -r requirements.txt

We also evaluate the robustness with Auto-Attack. It can be installed via following source code:

pip install git+https://github.com/fra31/auto-attack

We show sample usages in run.sh:

bash run.sh

For Tiny-ImageNet experiments, please prepare the dataset first under the path datasets/tiny-imagenet-200/.

For more detailed settings of different datasets, please refer to the supplementary material.

@inproceedings{li2022subspace,
  title={Subspace Adversarial Training},
  author={Li, Tao and Wu, Yingwen and Chen, Sizhe and Fang, Kun and Huang, Xiaolin},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={13409--13418},
  year={2022}
}