4 - FAU MoD Lecture: Image Reconstruction – The Dialectic of Modelling and Learning [ID:53677]
50 von 633 angezeigt

Okay, good afternoon to all of you.

Thank you for joining us to celebrate the spring.

So from the Mathematics of Data Center here at FAU, we try to run this series of dissemination

lectures delivered by distinguished scientists.

Today we have with us Martin Nuguder.

Thank you, Martin, for accepting the invitation.

I think Martin doesn't need much of an introduction.

So he's well known in the broad field of applied mathematics for his contributions in different

areas, including partial differential equations and medical analysis, but also imaging and

inverse problems.

Lately, he's interested in the interplay between the classical methods of computational mathematics

and machine learning.

His lecture today is precisely how they do it.

So the mathematics of image reconstruction, very much about imaging and inverse problems.

And his tension and mutual cooperation between classical continuous mechanics, modeling and

machine learning.

So thank you, Martin, for coming.

So Martin, I forgot to say, he has been a professor here for five years, until very

recently when he moved to D.C. in Hamburg as one of the research leaders there, together

with an appointment as a chair professor at the University of Hamburg.

Before that, he was a professor in Worcester.

And before that, I forgot.

So you can tell us maybe a little bit about yourself.

Before that, I was mainly in Austria and UCLA.

So UCLA, all nice places, a little bit in Italy.

But OK, yeah, thanks a lot for the invitation and the opportunity to speak back here.

I've done several lectures in this room, I think, for some years.

But then also we had to do some parts online during the Covid pandemic.

So it's good that we have these talks again in person, although I see there's a lot of

people popping into the Zoom as well.

Yeah, so I'm now since one year, almost one year at DESE, which is the Helmholtz Research

Center, which was the German accelerator somehow, the synchrotron.

So what I'm interested in there is mainly questions that arise from synchrotron radiation

and corresponding X-ray and laser imaging.

So we have this ring that you see partly here, which is PIPA 3, which produces high energy

X-ray.

The electron acceleration in this experiment, there are a lot of beam lines that people do

imaging, try to understand what they get.

We have this linear accelerator here that stops here in the flash building.

So there is the electron laser and then there's this another thing, European X-ray, which

goes a few kilometers to Schoenefeld in English with Goldstein.

So there's a lot of data produced there and the things that people want to do.

There will be even more data that will be produced hopefully in a few years.

We currently have applied to the Bundestag for PIPA 3 to PIPA 4, which is a 1.3 billion

euro project.

And that will be again the best light sources and gluton source in the world.

We are currently at number two, but people in China and Japan are building up, so we

will not keep that in the next years, but we want to stay kind of leading synchrotron

sources.

And what I'm mainly interested in is image reconstruction, which is sometimes the underestimated

Teil einer Videoserie :

Zugänglich über

Offener Zugang

Dauer

01:02:46 Min

Aufnahmedatum

2024-03-20

Hochgeladen am

2024-08-30 13:16:04

Sprache

en-US

Date: Wed. March 20, 2024
Event: FAU MoD Lecture
Event type: On-site / Online
Organized by: FAU MoD, the Research Center for Mathematics of Data at Friedrich-Alexander-Universität Erlangen-Nürnberg (Germany)

FAU MoD Lecture: Image Reconstruction – The Dialectic of Modelling and Learning

Speaker: Prof. Dr. Martin Burger
Affiliation: DESY, German Electron Synchrotron and Universität Hamburg

Abstract. In this talk we will discuss some current and future challenges in high-dimensional image reconstruction, which is based on the solution of large-scale inverse problems involving various uncertainties. While classical methods were purely based on physical models for forward operators and regularizations, modern machine learning techniques create the antithesis of data-driven approaches. We will discuss some pitfalls that machine learning can encounter in inverse problems and discuss opportunities for the synthesis of model- and data-driven approaches.

You can find more details of this FAU MoD lecture at:

https://mod.fau.eu/fau-mod-lecture-image-reconstruction-the-dialectic-of-modelling-and-learning/

 

 

 

Tags

Research Datenkommunikation Netzwerk Lecture FAU Lecture Series FAU MoD FAU MoD Lecture
Einbetten
Wordpress FAU Plugin
iFrame
Teilen