VillanDiffusion: A Unified Backdoor Attack Framework for Diffusion Models. (arXiv:2306.06874v2 [cs.CR] UPDATED)

Diffusion Models (DMs) are state-of-the-art generative models that learn a
reversible corruption process from iterative noise addition and denoising. They
are the backbone of many generative AI applications, such as text-to-image
conditional generation. However, recent studies have shown that basic
unconditional DMs (e.g., DDPM and DDIM) are vulnerable to backdoor injection, a
type of output manipulation attack triggered by a maliciously embedded pattern
at model input. This paper presents a unified backdoor attack framework
(VillanDiffusion) to expand the current scope …

addition applications art attack backdoor basic corruption diffusion models dms framework generative generative ai learn noise process state studies text vulnerable