Enhancing the Transferability of Targeted Attacks with Adversarial Perturbation Transform
The transferability of adversarial examples has been proven to be a potent tool for successful attacks on target models, even in challenging black-box environments. However, the majority of current research focuses on non-targeted attacks, making it arduous to enhance the transferability of targeted...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-09-01
|
| Series: | Electronics |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2079-9292/12/18/3895 |
| _version_ | 1797580473949487104 |
|---|---|
| author | Zhengjie Deng Wen Xiao Xiyan Li Shuqian He Yizhen Wang |
| author_facet | Zhengjie Deng Wen Xiao Xiyan Li Shuqian He Yizhen Wang |
| author_sort | Zhengjie Deng |
| collection | DOAJ |
| description | The transferability of adversarial examples has been proven to be a potent tool for successful attacks on target models, even in challenging black-box environments. However, the majority of current research focuses on non-targeted attacks, making it arduous to enhance the transferability of targeted attacks using traditional methods. This paper identifies a crucial issue in existing gradient iteration algorithms that generate adversarial perturbations in a fixed manner. These perturbations have a detrimental impact on subsequent gradient computations, resulting in instability of the update direction after momentum accumulation. Consequently, the transferability of adversarial examples is negatively affected. To overcome this issue, we propose an approach called Adversarial Perturbation Transform (APT) that introduces a transformation to the perturbations at each iteration. APT randomly samples clean patches from the original image and replaces the corresponding patches in the iterative output image. This transformed image is then used to compute the next momentum. In addition, APT could seamlessly integrate with other iterative gradient-based algorithms, incurring minimal additional computational overhead. Experimental results demonstrate that APT significantly enhances the transferability of targeted attacks when combined with traditional methods. Our approach achieves this improvement while maintaining computational efficiency. |
| first_indexed | 2024-03-10T22:50:31Z |
| format | Article |
| id | doaj.art-afaf5429fad04425ab7671be0b50d10f |
| institution | Directory Open Access Journal |
| issn | 2079-9292 |
| language | English |
| last_indexed | 2024-03-10T22:50:31Z |
| publishDate | 2023-09-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Electronics |
| spelling | doaj.art-afaf5429fad04425ab7671be0b50d10f2023-11-19T10:22:52ZengMDPI AGElectronics2079-92922023-09-011218389510.3390/electronics12183895Enhancing the Transferability of Targeted Attacks with Adversarial Perturbation TransformZhengjie Deng0Wen Xiao1Xiyan Li2Shuqian He3Yizhen Wang4School of Information Science and Technology, Hainan Normal University, Haikou 571158, ChinaSchool of Information Science and Technology, Hainan Normal University, Haikou 571158, ChinaSchool of Information Science and Technology, Hainan Normal University, Haikou 571158, ChinaSchool of Information Science and Technology, Hainan Normal University, Haikou 571158, ChinaSchool of Physics and Electronic Engineering, Hainan Normal University, Haikou 571158, ChinaThe transferability of adversarial examples has been proven to be a potent tool for successful attacks on target models, even in challenging black-box environments. However, the majority of current research focuses on non-targeted attacks, making it arduous to enhance the transferability of targeted attacks using traditional methods. This paper identifies a crucial issue in existing gradient iteration algorithms that generate adversarial perturbations in a fixed manner. These perturbations have a detrimental impact on subsequent gradient computations, resulting in instability of the update direction after momentum accumulation. Consequently, the transferability of adversarial examples is negatively affected. To overcome this issue, we propose an approach called Adversarial Perturbation Transform (APT) that introduces a transformation to the perturbations at each iteration. APT randomly samples clean patches from the original image and replaces the corresponding patches in the iterative output image. This transformed image is then used to compute the next momentum. In addition, APT could seamlessly integrate with other iterative gradient-based algorithms, incurring minimal additional computational overhead. Experimental results demonstrate that APT significantly enhances the transferability of targeted attacks when combined with traditional methods. Our approach achieves this improvement while maintaining computational efficiency.https://www.mdpi.com/2079-9292/12/18/3895adversarial examplestransferabilityperturbation transformtargeted attacks |
| spellingShingle | Zhengjie Deng Wen Xiao Xiyan Li Shuqian He Yizhen Wang Enhancing the Transferability of Targeted Attacks with Adversarial Perturbation Transform Electronics adversarial examples transferability perturbation transform targeted attacks |
| title | Enhancing the Transferability of Targeted Attacks with Adversarial Perturbation Transform |
| title_full | Enhancing the Transferability of Targeted Attacks with Adversarial Perturbation Transform |
| title_fullStr | Enhancing the Transferability of Targeted Attacks with Adversarial Perturbation Transform |
| title_full_unstemmed | Enhancing the Transferability of Targeted Attacks with Adversarial Perturbation Transform |
| title_short | Enhancing the Transferability of Targeted Attacks with Adversarial Perturbation Transform |
| title_sort | enhancing the transferability of targeted attacks with adversarial perturbation transform |
| topic | adversarial examples transferability perturbation transform targeted attacks |
| url | https://www.mdpi.com/2079-9292/12/18/3895 |
| work_keys_str_mv | AT zhengjiedeng enhancingthetransferabilityoftargetedattackswithadversarialperturbationtransform AT wenxiao enhancingthetransferabilityoftargetedattackswithadversarialperturbationtransform AT xiyanli enhancingthetransferabilityoftargetedattackswithadversarialperturbationtransform AT shuqianhe enhancingthetransferabilityoftargetedattackswithadversarialperturbationtransform AT yizhenwang enhancingthetransferabilityoftargetedattackswithadversarialperturbationtransform |