arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2403.13017cs.CRcs.AI
收藏

Impart:一种不可感知且有效的标签特定后门攻击

Impart: An Imperceptible and Effective Label-Specific Backdoor Attack

Tianjin University(天津大学);Nanyang Technological University(南洋理工大学)

Jingke Zhao, Zan Wang, Yongwei Wang, Lanjun Wang

AI总结:

提出名为Impart的新型不可感知后门攻击框架,通过代理模型生成与目标标签对齐的扰动,在攻击者无法访问受害者模型且all-to-all设置下显著提升攻击能力。

摘要:

后门攻击已被证明对现实安全关键场景构成严重威胁。尽管先前的工作能够实现较高的攻击成功率,但它们要么需要访问受害者模型,这在实践中可能显著降低其威胁;要么在隐蔽性上表现为视觉上可察觉。此外,在不同的中毒样本可能具有不同目标标签的场景(即 all-to-all 设置)下,攻击成功率仍有提升空间。在本研究中,我们提出了一种名为 Impart 的新型不可感知后门攻击框架,适用于攻击者无法访问受害者模型的场景。具体而言,为了增强 all-to-all 设置下的攻击能力,我们首先提出了一种标签特定攻击。与先前试图找到不可感知模式并将其作为中毒图像添加到源图像中的工作不同,我们随后提出通过一个代理模型生成在图像特征上与目标标签对齐的扰动。通过这种方式,生成的中毒图像被附加了关于目标类的知识,从而显著增强了攻击能力。

英文摘要:

Backdoor attacks have been shown to impose severe threats to real security-critical scenarios. Although previous works can achieve high attack success rates, they either require access to victim models which may significantly reduce their threats in practice, or perform visually noticeable in stealthiness. Besides, there is still room to improve the attack success rates in the scenario that different poisoned samples may have different target labels (a.k.a., the all-to-all setting). In this study, we propose a novel imperceptible backdoor attack framework, named Impart, in the scenario where the attacker has no access to the victim model. Specifically, in order to enhance the attack capability of the all-to-all setting, we first propose a label-specific attack. Different from previous works which try to find an imperceptible pattern and add it to the source image as the poisoned image, we then propose to generate perturbations that align with the target label in the image feature by a surrogate model. In this way, the generated poisoned images are attached with knowledge about the target class, which significantly enhances the attack capability.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑