The Institute for Data Science, Engineering, and Analytics (IDE+A), which is part of the THK-AI Research Cluster at TH Köln participated in the 35th Workshop Computational Intelligence (CI Workshop) held in Berlin. The workshop serves as a forum for academic research and industrial applications, focusing on the integration of AI methods into engineering, control systems, and material science.
Two contributions from the research group led by Prof. Dr. Thomas Bartz-Beielstein were accepted into the program and presented at the event.
Research on Explainability and Hybrid Modeling
Alexander Hinterleitner, a doctoral candidate at the institute, presented the paper „Tuning for Explainability: Incorporating XAI Consistency into Multi-Objective Hyperparameter Optimization“. This work was conducted in collaboration with industry partner Everllence SE. The research focuses on treating interpretability as a design objective by quantifying agreement among diverse feature attribution methods. Alexander’s doctoral research is supervised in cooperation with Prof. Dr. Oliver Niggemann from the Helmut-Schmidt-University (HSU) Hamburg. This cooperative supervision is based on a long-standing research cooperation between Prof. Dr. Bartz-Beielstein and Prof. Niggemann.
Additionally, Aleksandr Subbotin presented the paper „Physics-Informed Neural Networks for State Estimation Problem“. This study proposes a method integrating Physics-Informed Neural Networks (PINNs) with the Unscented Kalman Filter (UKF) to address state estimation in nonlinear dynamical systems. This doctoral project is being carried out within the framework of the Promotionskolleg NRW (PK NRW). Prof. Dr. Bartz-Beielstein served as a founding co-director of the PK NRW, which was established to confer independent doctoral degrees at universities of applied sciences in North Rhine-Westphalia.
Workshop Participants and Scope
The 35th CI Workshop was chaired by Prof. Horst Schulte (HTW Berlin). The program included keynote lectures on safety in learning control systems by Daniel Görges (TU Kaiserslautern-Landau) and on machine learning in material science by Michael Möckel (TH Aschaffenburg).
Contributions were presented by representatives from various institutions, including:
- Karlsruhe Institute of Technology (KIT),
- University of Kassel,
- HTW Berlin,
- Hochschule Fulda,
- Hochschule Bielefeld,
- TU Dortmund,
- Ruhr-Universität Bochum,
- Universität Siegen,
- FH Vorarlberg,
- University of Ljubljana, and the
- Bundesanstalt für Materialforschung und -prüfung (BAM), as well as from several industrial partners such as Robert Bosch GmbH and Everllence SE.
„The presented papers illustrate the application of AI methods within engineering contexts,“ stated Prof. Dr. Thomas Bartz-Beielstein. „The cooperative doctoral procedures, facilitated through the PK NRW and the cooperation with HSU Hamburg, enable the combination of academic research with specific industrial requirements.“
Contact:
Prof. Dr. Thomas Bartz-Beielstein, Research Cluster THK-AI, Institute for Data Science, Engineering, and Analytics (IDE+A), TH Köln, Steinmüllerallee 1, 51643 Gummersbach.