ErUM-Data Workshop on Inverse Problems

5 Dec 2023, 08:30 → 6 Dec 2023, 18:00 Europe/Berlin

Room E.0.11. (Campus Garching)

Due to the difficult travel conditions, the workshop is also being broadcasted online. Join here: https://rwth.zoom-x.de/j/67466745362 

This Workshop is hosted by the community organization DIG-UM with support from the BMBF-funded ErUM-Data-Hub.

The goal of the workshop is to get an overview on inverse problems in the ErUM-Data community, to discuss methods to solve them, and to foster exchange and cooperation between the ErUM-communities. In particular, the workshop is intended for early stage researchers.

Topics to be discussed include:

• Inverse problems in Research on Universe and Matter (ErUM)
• Bayesian reasoning
• Differential programming
• Generative models
• Score based models (incl. diffusion models)
• Dimensionality reduction
• ErUM strategy

 

Confirmed Speaker:

• Niels Thürey (TUM): Diffusion models
• Oliver Schulz (MPP): Differential and probabilistic programming
• Daniel Boschmann (EvalSpek-ML): Disassembling linear combinations into their constituent parts
• Philipp Frank (MPA, ErUM-IFT): Signal reconstruction for fields
• Frank Schreiber (Uni Tübingen, VIPR): Machine Learning for Surface Scattering
• Jonas Kreidelmeyer  (CFEL, 4D-KI-track): Multi-detector 3D reconstruction for Xray-Fluorescence Imaging

 

A registration fee of 80€ will be charged for the workshop. 

The call for abstracts ends on 27.10.2023.

Tuesday, 5 December

08:30 Coffee

Welcome

Session 1: Diffusion and State Space Models (Chair Vincent Eberle)

1 Invited Talk: Niels Thürey (TUM): Diffusion models

2 Out-of-Distribution Multi-set Generation with Context Extrapolation for Amortized Simulation and Inverse Problems

Speaker: Hosein Hashemi (BELLE (BELLE II Experiment))

11:20 Coffee

Session 1: Tomography (Chair Margret Westerkamp)

3 Invited Talk: Jonas Kreidelmeyer (CFEL, 4D-KI-track): Multi-detector 3D reconstruction for Xray-Fluorescence Imaging

Multi-detector 3D reconstruction for Xray-Fluorescence Imaging - Jonas Kreidelmeyer.pdf

4 KI4D4E: An AI-based framework for the visualization and evaluation of 4D tomography data for beamline end-users

Speaker: Tak Ming Wong (Helmholtz-Zentrum Hereon)

13:15 Lunch

Session 2 (Chair Margret Westerkamp): Photons and Neutrons

5 Invited Talk: Frank Schreiber - Machine Learning for Surface Scattering

Speaker: Frank Schreiber (Universität Tübingen)

6 Transformer based full parameter fitting of Neutron Scattering Experiments

Speaker: Paul Weimer (TUM, FRM II)

16:00 Coffee

Session 2 (Chair Margret Westerkamp): Photons and Neutrons

7 X-ray and neutron scattering data for machine learning of invertible neural networks

Speaker: Julian Eliah Heger (Technical University of Munich, TUM School of Natural Sciences, Department of Physics, Chair for Functional Materials)

231205_Heger_ErUM.pdf

8 Optimizing synchrotron radiation based microtomography experiments

Speaker: Thomas Jentschke (Hereon (Helmholtz-Zentrum Hereon))

9 Enabling online phase retrieval for in-situ and operando x-ray imaging

Speakers: Andre Lopes Marinho (FS-PS (Photon Science)), Johannes Hagemann (FS-PETRA (PETRA III))

Workshop Dinner

Wednesday, 6 December

08:00 Coffee

Session 3 (Chair Matteo Guardiani): Imaging

10 Invited Talk: Philipp Frank (MPA, ErUM-IFT): Signal reconstruction for fields

erum_ift_philipp_frank.pdf

11 Bayesian radio interferometric imaging

Speaker: Jakob Roth (Max Planck Institut for Astrophysics)

Resolve_Jakob_Roth.pdf

12 First spatio-spectral Bayesian imaging of SN1006 in Xray - Margret Westerkamp

11:00 Coffee

Session 4 (Chair Jakob Roth): Differential Programming and Inverse Problems

13 Invited Talk: Oliver Schulz (MPP): Differential and probabilistic programming

Speaker: Oliver Schulz (MPI Munich)

14 Unfolding Atmospheric Neutrino Spectra

Speaker: Tim Ruhe (TU Dortmund)

Inverse_Problems_Garching.pdf

13:00 Lunch

ErUM-Data Strategy Discussion

Convener: Thomas Kuhr (BELLE (BELLE II Experiment))

ErUM-Data-Strategy-Discussion.pdf

15:30 Coffee

Session 4 (Chair Jakob Roth): Recovering Component Relations

15 Invited Talk: Daniel Boschmann (EvalSpek-ML): Disassembling linear combinations into their constituent parts

Speaker: Daniel Boschmann (Helmut-Schmidt-Universität/ Bundeswehr Universität Hamburg)

ErUM-Data Workshop.pptx

16 Bayesian Causal Inference - Matteo Guardiani

BCI.pdf

Farewell

Inverse Problems.pdf