Seminar by SysCon: Event-triggered Learning and Similarity of Dynamical Systems
Fredrich Solowjow. Photo: private
- Date: 10 September 2025, 13:15–14:00
- Location: Ångström Laboratory, 100155, Theatrum Visuale
- Type: Seminar
- Lecturer: Friedrich Solowjow
- Web page
- Organiser: Department of Information Technology: Division of Systems and Control
- Contact person: Friedrich Solowjow
- Phone:
We are delighted to welcome Friedrich Solowjow, senior lecturer and deputy head of the Institute for Data Science in Mechanical Engineering at RWTH Aachen University in Germany.
With a PhD from the Max Planck Institute for Intelligent Systems and the University of Tübingen, Friedrich conducts research at the intersection of machine learning and control theory. Join us for an inspiring seminar where groundbreaking research meets practical application.
The seminar is open to all interested and will be held in English.
Abstract:
In learning-based control, data is inherently non-iid due to the closed-loop interaction between controller and system, with excitation and exploration directly influencing the underlying distributions. Thus, the problem is fundamentally different from classical machine learning. An additional challenge arises from the tension between stabilizing the system and collecting informative data: effective controllers suppress deviations from a setpoint, but this often limits the information available for learning. Furthermore, changing dynamics from new tasks, varying loads, or altered environments require controllers to adapt, raising the key question: when should learning be triggered?
I address this by introducing event-triggered learning (ETL), a principled framework for deciding when to update models based on statistical evidence. At its core, ETL employs kernel-based similarity tests to detect significant changes in dynamical systems, enabling timely and data-efficient adaptation. In the end, I will talk about recent medical and industrial applications.
Short Bio:
Friedrich is a senior lecturer and deputy head of the Institute for Data Science in Mechanical Engineering at RWTH Aachen University in Germany. He received his PhD from the Max Planck Institute for Intelligent Systems and the University of Tübingen. His research focuses on the intersection of machine learning and control theory.