Modern Regularisation


The availability of expressive regularizers is a very important component in solving ill-posed inverse problems in imaging. In recent years, hand-designed regularizers have been gradually replaced by data-driven ones. Provided sufficient training data is available, it is nowadays possible to learn tailored regularizers for a certain problem class. This usually leads to a huge increase in reconstruction quality, but the learned regularizers are usually much harder to analyse and it is much harder to give guarantees on convergence behaviour, generalisation ability, or reconstruction error. In this session, we will present and discuss the latest methods, techniques and applications in this cutting-edge field of research.

Organisers: Thomas Pock (Chair) & Philip Schniter (Committee member)

Session Schedule

17:30 - 17:50 Invited talk -
An Introduction to Modern Regularisation
Thomas Pock
17:50 - 18:10 Invited talk -
Solving inverse problems in imaging via learned regularizers and learned image features
Stancey Levine
18:10 - 18:30 Invited talk -
Risk Control for Online Learning Models
Yaniv Romano
18:30 - 18:50 Invited talk -
Generative regularization approaches in imaging – a function space perspective
Martin Holler
18:50 - 19:10 Invited talk -
Critical point regularization with non-convex regularizers
Markus Haltmeier
19:10 - 20:30 Invited poster -
Adaptive Sparse Phase Recovery for Efficient Reconstruction of Holographic Lens-Free Images
René Vidal
19:10 - 20:30 Contributed poster -
Statistical Sparsity, Non-Convexity, and Smoothness, All in One Place: The Cauchy Penalty
Alin Achim
19:10 - 20:30 Contributed poster -
Modified Loss Functions for Improved Plug-and-Play Methods
Danica B. Fliss*
19:10 - 20:30 Committee member poster -
Deep regularization via bi-level optimization with adversarial regularization
Philip Schniter
19:10 - 20:30 Committee member poster -
Deep network series for large-scale high-dynamic range imaging
Yves Wiaux