Self-Adaptive Perturbation Radii for Adversarial Training

Document Type

Conference Proceeding

Publication Title

Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining

Abstract

Adversarial training has been shown to be the most popular and effective technique to protect models from imperceptible adversarial samples. Despite its success, it also accompanies the significant performance degeneration to clean data. To achieve a good performance on both clean and adversarial samples, the main effort is searching for an adaptive perturbation radius for each training sample. However, this method suffers from a conflict between exact searching and computational overhead. To address this conflict, in this paper, firstly we show the superiority of adaptive perturbation radii on the accuracy and robustness respectively. Then we propose our novel self-adaptive adjustment framework for perturbation radii without tedious searching. We also discuss this framework on both deep neural networks (DNNs) and kernel support vector machines (SVMs). Finally, extensive experimental results show that our framework can improve adversarial robustness without compromising the natural generalization. It is also competitive with existing searching strategies in terms of running time.

First Page

2570

Last Page

2581

DOI

10.1145/3580305.3599495

Publication Date

8-4-2023

Keywords

adversarial training, self-adaptive perturbation radii

Comments

IR conditions: non-described

Share

COinS