Harnessing Low Dimensionality in Diffusion Models: From Theory to Practice: Lecture III: Diffusion Inverse Solvers for Scientific Applications

Qing Qu, Yuxin Chen, Liyue Shen

International Conference on Machine Learning 2025 · Tutorial

This article delves into the third lecture of a comprehensive tutorial on diffusion models, focusing specifically on their application as inverse solvers for scientific problems. Presented by Liyue Shen, the talk bridges the theoretical foundations of diffusion models, previously covered in the first two lectures by Dr. Qing Qu and Dr. Yuxin Chen, with their profound practical implications. The core theme revolves around how the powerful data distribution priors learned by diffusion models can be leveraged to tackle challenging ill-posed inverse problems prevalent across various scientific and engineering domains.

AI review

This is Lecture III of a tutorial series on diffusion models, delivered by Liyue Shen, covering the application of diffusion model priors to scientific inverse problems — primarily medical imaging (CT, MRI). The lecture surveys the problem landscape, identifies three organizing challenges (efficiency, generalization, controllability), and presents several of Shen's own contributions (Diffusion Blend, Resample, CCS) as responses to those challenges. The work is competent and practically motivated, and the framing around posterior sampling is conceptually clean. However, as reviewed here — and…