2/4/2025 | 11:00 AM - 12:20 PM | Regency C

Diffusion posterior sampling with inaccurate priors or physics

Author(s)

Weimin Bai
Yifei Wang
Siyi Chen
Wenzheng Chen
He Sun

Abstract

Deep generative models have revolutionized the field of computational due to their exceptional ability to model complex prior distributions. However, their reliance on extensive, clean datasets for learning or accurate physics model for inference limits their practical use where clean data is scarce or physics model is inaccurate, respectively. In this paper, we propose an expectation-maximization (EM) approach for diffusion posterior sampling with inaccurate priors or physics. Our method alternates between reconstructing clean medical images from corrupted clinic data using a known diffusion prior and physical model (E-step) and refining the prior or physical model based on these reconstructions (M-step). This iterative process gradually guides the learned image prior or physical model to converge to true clean distribution. We validate our method through extensive experiments on diverse computational imaging tasks, including inpainting, deblurring and denoising, achieving new state-of-the-art performance.