Saved in:
Bibliographic Details
Main Authors: Yuan, Zheng, Zhang, Jie, Jiang, Zhaoyan, Li, Liangliang, Shan, Shiguang
Format: Preprint
Published: 2021
Subjects:
Online Access:https://arxiv.org/abs/2111.13841
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913244855140352
author Yuan, Zheng
Zhang, Jie
Jiang, Zhaoyan
Li, Liangliang
Shan, Shiguang
author_facet Yuan, Zheng
Zhang, Jie
Jiang, Zhaoyan
Li, Liangliang
Shan, Shiguang
contents In recent years, the security of deep learning models achieves more and more attentions with the rapid development of neural networks, which are vulnerable to adversarial examples. Almost all existing gradient-based attack methods use the sign function in the generation to meet the requirement of perturbation budget on $L_\infty$ norm. However, we find that the sign function may be improper for generating adversarial examples since it modifies the exact gradient direction. Instead of using the sign function, we propose to directly utilize the exact gradient direction with a scaling factor for generating adversarial perturbations, which improves the attack success rates of adversarial examples even with fewer perturbations. At the same time, we also theoretically prove that this method can achieve better black-box transferability. Moreover, considering that the best scaling factor varies across different images, we propose an adaptive scaling factor generator to seek an appropriate scaling factor for each image, which avoids the computational cost for manually searching the scaling factor. Our method can be integrated with almost all existing gradient-based attack methods to further improve their attack success rates. Extensive experiments on the CIFAR10 and ImageNet datasets show that our method exhibits higher transferability and outperforms the state-of-the-art methods.
format Preprint
id arxiv_https___arxiv_org_abs_2111_13841
institution arXiv
publishDate 2021
record_format arxiv
spellingShingle Adaptive Perturbation for Adversarial Attack
Yuan, Zheng
Zhang, Jie
Jiang, Zhaoyan
Li, Liangliang
Shan, Shiguang
Computer Vision and Pattern Recognition
In recent years, the security of deep learning models achieves more and more attentions with the rapid development of neural networks, which are vulnerable to adversarial examples. Almost all existing gradient-based attack methods use the sign function in the generation to meet the requirement of perturbation budget on $L_\infty$ norm. However, we find that the sign function may be improper for generating adversarial examples since it modifies the exact gradient direction. Instead of using the sign function, we propose to directly utilize the exact gradient direction with a scaling factor for generating adversarial perturbations, which improves the attack success rates of adversarial examples even with fewer perturbations. At the same time, we also theoretically prove that this method can achieve better black-box transferability. Moreover, considering that the best scaling factor varies across different images, we propose an adaptive scaling factor generator to seek an appropriate scaling factor for each image, which avoids the computational cost for manually searching the scaling factor. Our method can be integrated with almost all existing gradient-based attack methods to further improve their attack success rates. Extensive experiments on the CIFAR10 and ImageNet datasets show that our method exhibits higher transferability and outperforms the state-of-the-art methods.
title Adaptive Perturbation for Adversarial Attack
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2111.13841