This tutorial discusses the challenges and opportunities of using modern AI for inverse problems and scientific applications more broadly. In particular I will discuss an emerging literature on deep learning for inverse problems that have been very successful for a variety of image and signal recovery and restoration tasks ranging from denoising and MR reconstruction to nano-scale imaging (with a special focus on phase retrieval). This tutorial will focus on the two most successful deep learning paradigms in this domain which are end-to-end training methods and diffusion models. I will review existing literature and discuss various challenges and opportunities in this emerging discipline.
Slide deck can be found below: