Summer School 

Recap

From 4 October to 8 October, the first edition of the SPP 2298 Summer School took place at TUM Science Center in Raitenhaslach. The event aimed to present key topics relating to the mathematical foundations of deep learning and to provide a platform for young researchers in the field to network. The name was by no means misleading, as participants were able to spend four sunny days at the historic monastery complex at the foot of the world’s longest castle. The Summer School centred on three two-part lectures, which first provided an introduction to the topic and then, building on this, explored various aspects of modern research in the field. This enabled every participant, even without any specific prior knowledge, to understand the fundamental questions in this field and to identify links to other areas. These were complemented by open-ended working sessions during which participants worked together with the lecturers on short exercises related to the topics and continued to exchange ideas. We would like to thank the lecturers Johannes Schmidt-Hieber, Lars Grüne and Christian Wald for their wonderfully prepared lectures and exercises, which perfectly captured the spirit of the event. In addition, there was a special lecture by Tim Roith, who stepped in at short notice for Leo Schwinn, to mark the end of the four days. The supporting programme was rounded off by a poster session in the monastery courtyard at the start of the summer school, a social dinner in Burghausen, and various group activities around the site. The participants were quickly able to make new contacts here and, over the four days, gain an insight into a number of new mathematical concepts which will hopefully assist them in their future research. A big thank you goes to the SPP 2298 for funding the Summer School and to all the participants who enriched it through their active involvement.

About the Summer School

The aim is to make young researchers familiar with a broad range of theoretical concepts that are relevant in the field of deep learning. At the same time, it seeks to strengthen exchange and foster collaboration among the projects funded within the SPP.

Participation does not require prior knowledge of the covered topics. In addition to the lecture programme, participants will have the opportunity to present their research in a poster session.

The summer school is funded by the Deutsche Forschungsgemeinschaft (DFG) within the Priority Programme 2298 “Theoretical Foundations of Deep Learning".

Organizers

• Christopher Bülte (LMU Munich)

• Samira Kabri (DESY Hamburg)

• Sara-Viola Kuntz (Technical University of Munich)

• Lucas Schmitt (University of Würzburg)

• Mario Sperl (University of Bayreuth)

For any questions, please send an e-mail to spp2298@math.lmu.de