Diffusion models as plug-and-play priors
This approach allows adapting models to new domains and tasks, potentially leading to new algorithms for combinatorial optimization, but it appears incremental as it builds on existing diffusion model frameworks.
The paper tackles the problem of inferring high-dimensional data using a prior from a denoising diffusion generative model and an auxiliary differentiable constraint, enabling diffusion models to serve as plug-and-play modules for tasks like conditional generation or image segmentation.
We consider the problem of inferring high-dimensional data $\mathbf{x}$ in a model that consists of a prior $p(\mathbf{x})$ and an auxiliary differentiable constraint $c(\mathbf{x},\mathbf{y})$ on $x$ given some additional information $\mathbf{y}$. 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. The possibility of such inference turns diffusion models into plug-and-play modules, thereby allowing a range of potential applications in adapting models to new domains and tasks, such as conditional generation or image segmentation. The structure of diffusion models allows us to perform approximate inference by iterating differentiation through the fixed denoising network enriched with different amounts of noise at each step. Considering many noised versions of $\mathbf{x}$ in evaluation of its fitness is a novel search mechanism that may lead to new algorithms for solving combinatorial optimization problems.