IUQ Workshop 2022

Workshop: Imaging With Uncertainty Quantification (IUQ22), September 2022

A 3-day workshop on computational uncertainty quantification for inverse problems,
preceded by a 1-day training course on our software CUQIpy

Training course: Sept. 26

Workshop: Sept. 27-29, 2022

Held at conference center Konventum in Helsingør (Elsinore), north of Copenhagen.


Arranged by the research project CUQI.


Imaging applications are everywhere in science and technology, and often there is a need for assessing the uncertainty of the reconstructions due to measurement noise, model errors, etc. Hence, we see an increasing interest in performing uncertainty quantification (UQ) for imaging applications, and for making such methods readily useful in applications.

This workshop aims at bringing together specialists in UQ for imaging, and we invite talks that cover various aspects related to the development of theory, methodology and software. We also invite talks about interesting applications of UQ in imaging. The goal is to stimulate networking and collaboration between researchers and students in these areas.

Before the workshop, we arrange a 1-day short course devoted to the Python software CUQIpy that we are currently developing for modeling and computations related to UQ for imaging. The short course is primarily aimed at PhD students and postdocs having a rudimentary familiarity with Python.

The workshop and training course are part of the activities in the research project CUQI, Computational Uncertainty Quantification for Inverse problems, funded by Villum Fonden (the Villum Foundation). We also acknowledge financial support from the TIRCA Foundation and Otto Mønsteds Fond (the Otto Mønsted Foundation).
   

Getting to Konventum. There is a direct train from Copenhagen Airport to Helsingør Station once an hour - you can use the online travel guide "rejseplanen" to find connections. The conference center Konventum is located 3 km from Helsingør Station, and you can catch a taxi at the station (or call +45 48484848 for Helsingør Taxa). The taxi will cost 150-200 DKK.

Invited Speakers

A number of invited speakers will give talks related to the themes of the workshop:

  • Assoc. Prof. Thomas Mejer Hansen, Aarhus University, homepage
  • Professor Bangti Jin, Chinese University of Hong Kong, homepage
  • Assistant Prof. Hanne Kekkonen, TU Delft, homepage
  • Dr. Alexander Litvinenko, RWTH Aachen University, homepage
  • Assoc. Prof. Marcelo Pereyra, Heriot-Watt University, Edinburgh homepage
  • Professor Aku Seppänen, University of Eastern Finland homepage
  • Professor Tanja Tarvainen, University of Eastern Finland, homepage
  • Research Professor Faouzi Triki, Grenoble-Alpes University, homepage

The Training Course

The training course is an opportunity to learn about our software platform CUQIpy for modeling and computations with UQ for inverse problems. The goal of this software is to provide an easy-to-use platform that allows non-experts in UQ and Bayesian inversion to perform UQ analysis of inverse problems. At the same time, the software allows expert users full control of the models and methods.

After a brief introduction to UQ, we demonstrate the main features of the software via examples and hands-on exercises using the built-in use-cases. Then we describe more advanced features, and we demonstrate how to work with one's own inverse problem in the CUQIpy software.

Please bring a laptop.

Program

This is the final version of the program. A pdf file with the abstracts is available HERE.

Sunday, Sept. 25
19:00 DINNER
Monday, Sept. 26 Training course
09:00 - 10:30 Part 1
10:30 - 11:00 COFFEE BREAK
11:00 - 13:00 Part 2
13:00 - 14:00 LUNCH
14:00 - 15:00 Part 3
15:00 - 15:30 COFFEE BREAK
15:30 - 17:00 Part 4
19:00 DINNER
Tuesday, Sept. 27
09:00 - 09:15 Introduction
09:15 - 10:15 Marcelo Pereyra An introduction to Bayesian imaging with data-driven priors encoded by neural networks
10:15 - 10:45 Rémi Laumont Bayesian imaging using Plug & Play priors: when Langevin meets Tweedie
10:45 - 11:15 Coffee break
11:15 - 12:15 Hanne Kekkonen Random tree Besov priors
12:15 - 12:45 Yiqiu Dong The horseshoe prior for edge-preserving Bayesian inversion
13:00 - 14:00 LUNCH
14:00 - 15:00 Thomas Mejer Hansen Probabilistic integration of (geo)-information
15:00 - 15:30 Coffee break
15:30 - 16:00 Jakob S. Jørgensen CUQIpy: A new Pyhthon platform for computational uncertainty quantifiation for inverse problems
16:00 - 16:30 Johannes Hertrich The Power of Patches for Training Normalizing Flows
16:30 - 17:30 POSTER SESSION
  Andreas Horst, Elias Vänskä, Eva Havelková, Kenneth Scheel, Puyuan Mi, Silja L. Christensen,
17:30 - 18:30 Guided art tour at Konventum (for those who registered)
19:00 DINNER
Wednesday, Sept.28
09:00 - 10:00 Aku Seppänen Recovery from modelling errors in non-stationary inverse problems
10:00 - 10:30 Fabian Parzer Uncertainty-aware blob detection in astronomical imaging
10:30 - 11:00 Coffee break
11:00 - 12:00 Faouzi Triki Inverse moving point source problem for the wave equation
12:00 - 12:30 Remo Kretschmann Optimal and Bayesian hypothesis testing in statistical inverse problems
12:30 - 13:00 Mirza Karamehmedović Forward Uncertainty Quantification for the Helmholtz Equation
13:00 - 14:00 LUNCH
14:00 Bus departure for excursion to Kronborg (open till 17:00 - for those who registered)
(No bus back to Konventum!)
19:00 CONFERENCE DINNER at Konventum
Thursday, Sept. 29
09:00 - 10:00 Bangti Jin Deep Image Prior of Inverse Problems: Acceleration and Probability Treatment
10:00 - 10:30 Babak Maboudi Afkham Uncertainty quantification of inclusion boundaries in the context of X-ray tomography
10:30 - 11:00 Coffee break
11:00 - 12:00 Alexander Litvinenko Weakly Supervised Regression on Uncertain Datasets
12:00 - 12:30 Kim Knudsen Sound speed uncertainty in Acousto-Electric Tomography
12:30 - 13:00 Jasper Everink Sparse Bayesian Inference with Regularized Gaussians
13:00 - 14:00 LUNCH
14:00 - 15:00 Tanja Tarvainen Bayesian Approach to Quantitative Photoacoustic Tomography
15:00 - 15:30 Coffee break
15:30 - 16:00 Charlesquin Mbakam Empirical Bayesian estimation for semi-blind inverse problems: Application to image deblurring with total variation regularization
16:00 End of workshop

Last updated by Per Christian Hansen, September 22, 2022.