arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2210.14404cs.LGcs.CRcs.CV
收藏

利用流形假设的对抗净化

Adversarial Purification with the Manifold Hypothesis

GE Research(通用电气研究院);Australian National University(澳大利亚国立大学)

Zhaoyuan Yang, Zhiwei Xu, Jing Zhang, Richard Hartley, Peter Tu

AI总结:

本文基于流形假设提出对抗鲁棒性框架,结合流形学习与变分推断开发对抗净化方法,无需对抗训练即可防御对抗样本,并可作为变分自编码器的测试时防御。

摘要:

在这项工作中,我们利用流形假设构建了一个新的对抗鲁棒性框架。该框架为防御对抗样本提供了充分条件。我们基于此框架开发了一种对抗净化方法。我们的方法将流形学习与变分推断相结合,无需昂贵的对抗训练即可提供对抗鲁棒性。实验表明,即使攻击者知晓防御的存在,我们的方法也能提供对抗鲁棒性。此外,我们的方法还可作为变分自编码器的测试时防御机制。

英文摘要:

In this work, we formulate a novel framework for adversarial robustness using the manifold hypothesis. This framework provides sufficient conditions for defending against adversarial examples. We develop an adversarial purification method with this framework. Our method combines manifold learning with variational inference to provide adversarial robustness without the need for expensive adversarial training. Experimentally, our approach can provide adversarial robustness even if attackers are aware of the existence of the defense. In addition, our method can also serve as a test-time defense mechanism for variational autoencoders.

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。

↑