Date: Wed. May 15, 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: Using system knowledge for improved sample efficiency in data-driven modeling and control of complex technical systems

Speaker: Prof. Dr. Sebastian Peitz
Affiliation: Universität Paderborn (Germany)

Abstract. Modern technical systems such as autonomous vehicles, the electric grid or nuclear fusion reactors are extremely complex, which requires powerful techniques for predicting or controlling their behavior. As in almost all areas of science as well as our daily lives, machine learning has had a huge impact on the area of modeling and control of technical systems in recent years. However, the complexity of these systems renders the learning very data-hungry. The aim of this talk is thus to discuss different approaches to leverage system knowledge – and in particular symmetries – such that we can significantly improve the sample efficiency. Our discussion ranges from learning the dynamics from data to reinforcement learning. We will emphasize the benefits of exploiting knowledge using various examples from fluid mechanics.

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

https://mod.fau.eu/fau-mod-lecture-using-system-knowledge-for-improved-sample-efficiency-in-data-driven-modeling-and-control-of-complex-technical-systems/