Two papers from the THK-AI members were accepted at the 35. Workshop Computational Intelligence in Berline, November 2025: „Physics-Informed Neural Networks for State Estimation Problems“ and „Tuning for Explainability: Incorporating XAI Consistency into Multi-Objective Hyperparameter Optimization“. They will be presented at the conference and included as online-publications in the conference proceedings (Open Access in Karlsruher Institut für Technologie (KIT) Verlag).
The first paper is written by Aleksandr Subbotin and Thomas Bartz-Beielstein. It highlights the potential of blending machine learning with classical filtering techniques.
The second paper is based on a cooperation with Everllence SE (https://www.everllence.com) and written by Alexander Hinterleitner (TH Köln), Christoph Leitenmeier (Everllence SE), Sebastian Spengler (Everllence SE) and Thomas Bartz-Beielstein (TH Köln). This work extends XAI (Explainable Artificial Intelligence) by incorporating new measures and methodologies.
