Robust Diffusion Models for Adversarial Purification

arXiv:2403.16067v1 Announce Type: new
Abstract: Diffusion models (DMs) based adversarial purification (AP) has shown to be the most powerful alternative to adversarial training (AT). However, these methods neglect the fact that pre-trained diffusion models themselves are not robust to adversarial attacks as well. Additionally, the diffusion process can easily destroy semantic information and generate a high quality image but totally different from the original input image after the reverse process, leading to degraded standard accuracy. To overcome these issues, a …

abstract adversarial adversarial attacks adversarial training arxiv attacks cs.ai cs.cv diffusion diffusion models generate however information process robust semantic training type