DFG funding: At the interface between machine learning and control theory

Enlarge

In its most recent round of approvals, the German Research Foundation (DFG) decided to fund the new Research Training Group (RTG) "Machine Learning and Control Theory: Exploring Synergies, Complementarities and Mutual Benefits" (METEOR) at the LMU; the application was initiated jointly with the Technical University of Munich (TUM). METEOR will be funded for an initial period of five years starting in spring 2026. The RTG Team includes representatives from the fields of machine learning, control theory and applied mathematics. The Spokesperson is Professor Eyke Hüllermeier from LMU, co-spokesperson is Professor Sandra Hirche from TUM.

In light of the current challenges in the field of artificial intelligence (AI), METEOR is committed to combining machine learning (ML) and control theory (CT), as well as enhancing the collaboration between these two disciplines. Despite common interests and methods, the two fields have developed different languages and cultures, largely independently of each other. The combination of the learning-centered, data-driven approach of ML and the primarily model-based perspective of CT appears highly promising.

Interdisciplinary workshops and annual hackathons

Such a combination requires scientists with expertise on both sides. By combining cutting-edge research and comprehensive interdisciplinary training, METEOR will produce such a new generation of researchers at the intersection of ML and CT.

Specially designed lectures and research-oriented seminars are intended to create a common language and a basic understanding between the two disciplines. Innovative measures such as interdisciplinary workshops and annual hackathons build a further bridge between ML and CT and provide practical experience. In order to prepare graduates for a successful career in science and industry, the program is rounded off with targeted soft skills courses and international research stays.

“Foundations for a symbiotic relationship“

The research program of METEOR focuses on two main directions. On the one hand, it investigates how ML can support the data-driven design of robust control for complex, safety-critical applications ("ML for CT"). On the other hand, how concepts and methods of CT can contribute to the improvement of ML algorithms ("CT for ML"). Both directions are approached from the perspective of complex dynamical systems that form a common mathematical framework.

"The combination of data-driven methods from machine learning and model-centric approaches from control theory could be the key to creating next-generation AI systems. These systems should combine properties such as adaptivity, safety, robustness and explainability,” says METEOR-Spokesperson Eyke Hüllermeier. “In METEOR, we want to lay the foundations for this symbiotic relationship by opening up previously unexplored scientific territory between the disciplines, both in teaching and research."