Der zentrale Ansprechpartner für KI an der TH Köln.
A collaborative research paper by scientists from THK-AI Research Cluster, Toyota Research Institute, and Toyota Gazoo Racing Europe has been accepted at the prestigious NeurIPS 2025 conference. The paper, titled „From Faults to Features: Pretraining to Learn Robust Representations against Sensor Failures,“ introduces a novel pretraining approach to enhance the reliability of AI models in safety-critical applications like autonomous driving.
Der Podcast der TH Köln – Staffel 3 Episode 6: Keine Science-Fiction mehr: Roboter in der Pflege – Prof. Dr. … Mehr
Two papers from THK-AI members were accepted for presentation at the 35th Workshop on Computational Intelligence in Berlin, November 2025, and will be published as open access in the KIT Verlag proceedings. The first paper, by Aleksandr Subbotin and Thomas Bartz-Beielstein, explores combining machine learning with classical filtering methods. The second paper, a collaboration with Everllence SE, introduces new measures to enhance explainability in artificial intelligence.
Richard Schulz presented his talk today at the ACM GECCO 2025 in Malaga. Here are some impressions from his talk.
Prof. Bartz-Beielstein’s talk „Surrogate Model-Based Multi-Objective Optimization Using DesirabilityFunctions“ was presented at ACM Genetic and Evolutionary Computation Conference in Malaga. … Mehr
Gary Marcus claims that the practical demands of building capable AI systems have exposed fundamental limitations in pure deep learning … Mehr
The title of his talks is: „Surrogate Model-Based Multi-Objective Optimization UsingDesirability Functions“ Abstract: The desirability function approach is an established … Mehr
Der Antrag “KI-Nachwuchs@FH 2-2021: TH Köln – Künstliche Intelligenz plus (THK-KIplus)” wurde am 4.4.2023 bewilligt (Förderkennzeichen 13FH007KI2). Als Projektförderung wurde … Mehr
