Introduction

Inverse problems are ubiquitous in imaging sciences. Modern imaging methods rely predominantly on three mathematical and computational frameworks to formulate and solve imaging inverse problems: mathematical analysis, Bayesian statistics, and machine learning. These frameworks have complementary strengths and drawbacks in terms of the inferences they can support, their modularity and flexibility, their accuracy and computational efficiency, and the theoretical guarantees that they afford to practitioners.

This 3-day workshop focused on interactions between Bayesian statistics, machine learning, and applied analysis, with special attention (but not limited) to imaging inverse problems that are blind or semi-blind.  The aim was to bring together a group of world-leading experts and rising early career researchers to discuss recent developments in these fields and open challenges, with a focus on ideas that develop at the fertile interface where the three frameworks meet.

The workshop is supported by the UKRI EPSRC projects “Bayesian model selection and calibration for computational imaging” (EP/T007346/1) and “Learned Exascale Computational Imaging” (EP/W007673/1, EP/W007681/1).