Diffusionmodelsasplug-and-playpriors

We consider the problem of inferring high-dimensional datax in a model that consists of a priorp(x) and an auxiliary differentiable constraintc(x,y) on x given some additional informationy. In this paper, the prior is an independently trained denoising diffusion generative model. The auxiliary constraint is expected to have a differentiable form, but can come from diverse sources.