From Basics to Brilliance: Die erste ErUM-Data-Hub Deep Learning Schule „Basic Concepts”
Together we look back on an intensive, exciting and highly educational summer school.
The 5 days at Landhaus Nordhelle flew by for the more than 50 participants and the ErUM Data Hub team. The group of students, PhD students and post-docs arrived in Meinerzhagen on Monday, August 8. After the first dinner together, Dennis Noll (RWTH Aachen University) introduced the basics of Deep Learning with an evening lecture on Neural Network Building Blocks. The first thematic block was continued the next morning, after everyone had fortified themselves at the extensive breakfast buffet. In addition to teaching theoretical content, the school's goal was to give the participants the opportunity to apply their newly learned knowledge extensively and under supervision. Thus, a hands-on session on the topic of Neural Network Building Blocks was already held on the first morning together. Lunch and the subsequent break were used to socialize and to gather new energy for the afternoon in the picturesque landscape of the Sauerland.
The lecture and the practical part on the second topic block "Mastering Model Building" were designed by Dr. Marcel Rieger (University of Hamburg) and filled the Tuesday afternoon. The busy days at Landhaus Nordhelle were always rounded off by dinner and came to a relaxing end on the terrace with a beautiful view, lively conversations and refreshing drinks. Jun.-Prof. Judith Reindl (University of the Federal Armed Forces Munich) taught on Wednesday about Convolutional Neural Networks. A 90-minute hands-on session was also devoted to this topic. On Thursday, the group dove into Recurrent Neural Networks with Dr. Nikolai Hartmann (Ludwig-Maximilians-Universität München). The final module, Introspection of Neural Networks, was kicked off by Dr. Caroline Heneka (University of Hamburg) on Thursday evening and concluded with a final practice session on Friday morning. After a review of the productive week, the school ended with a joint lunch.
The hands-on sessions were actively supported by the tutors Lars Sowa (KIT), Bogdan Wiederspan (University of Hamburg) and Boyang Yu (LMU Munich). During the school, participants were encouraged to participate and were given the opportunity to act as session chairs, introduce the speakers and open the lectures.
The group from a total of 7 ErUM communities (KAT, KET, KfB, KFS, KFSI, KHuK and RDS) was highly interested and motivated and created an intensive working atmosphere through their active participation in the lectures and hands-on sessions. Despite bright sunshine, the work was concentrated. The participants also used the time beyond the event program to ask questions and discuss the contents among themselves and with the speakers.
Overall, the feedback on the school was consistently positive and illustrates once again that the demand for teaching Deep Learning is high in the ErUM communities. We look forward to offering another school on "Advanced Deep Learning Concepts" already in November 2022.
Thanks to all who attended!
We would like to thank the invited speakers and tutors for their great commitment, the Federal Ministry of Education and Research (BMBF) for the financial support and the participants, who contributed significantly to the success of the school through their active cooperation.