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.
@article{arxiv.2210.14404,
title = {Adversarial Purification with the Manifold Hypothesis},
author = {Zhaoyuan Yang and Zhiwei Xu and Jing Zhang and Richard Hartley and Peter Tu},
journal= {arXiv preprint arXiv:2210.14404},
year = {2023}
}
Comments
Extended version of paper accepted at AAAI 2024 with supplementary materials