Lectures on
"Learning for solving inverse problems"
14-15 november 2019

These two days will be dedicated to lectures and hands-on sessions in Python given by Peter Maass and Daniel Otero on learning for solving inverse problems.

Registration is free but mandatory.
Registration deadline: November, 7th.
Registration is closed.

For afternoon hands-on sessions, you need

• The installation of Python 3 the PyTorch (https://pytorch.org/) framework
• A google account (we are going to use Google Colab: https://colab.research.google.com/)

Everything is available on GitHub here.

Frumam, St Charles
Aix-Marseille Université
3, place Victor Hugo - MARSEILLE Cedex 03

The program is available here.

Briefly :

Thursday, November, 14th
8h45 Welcome
9h00-12h15 Introduction and motivation, ISTA and LISTA, Neural Networks for trivial ill-posed inverse problems, Deep Image Prior and mathematical aspects
12h15-13h45 Lunch
13h45-17h00 Applications on Magnetic Particle Imaging (MPI); Hands-on session


Friday, November, 15th
8h45 Welcome
9h00-12h15 More advanced topics (depending on how much we covered the first day)
12h15-13h45 Lunch
13h45-17h00 Applications on Computed Tomography (CT), Hands-on session

Lecture by Peter Maass [Slides].
Lecture by Daniel Otero [Introdution slides], [Slides].

Peter Maass , Universität Bremen.
Daniel Otero , Universität Bremen.

Caroline Chaux.
Bruno Torrésani.