Research Projects
Research projects I am or have been involved in as principal investigator. See also the work group page and the university research profile.
Active
CoFill — Intelligent Circular Economy Through Advanced AI
AI-driven data analysis and prediction models to improve efficiency and sustainability in the circular economy.
JetsKI — Waterjet Cutting with Smart AI Modeling
Applying artificial intelligence to optimize waterjet machining of difficult-to-machine materials like high-strength ceramics.
TrAIber.NRW — Transformation der automotiven Industrie in der bergischen Region
Supporting automotive suppliers in the Bergisches Land region in coping with drivetrain transitions via digitalization and AI-driven innovations.
Completed
Digitales Mentoring
Developing AI-supported tools for the study entry phase: an adaptive mathematics practice environment that adjusts to individual student abilities, Educational Data Mining to identify dropout risks and success factors, and a Learning Analytics acceptance study — across three universities of applied sciences.
SharKI — Shared Tasks als innovativer Ansatz zur Implementierung von KI- und Big-Data-basierten Anwendungen in der Hochschullandschaft
Developing shared tasks — friendly scientific competitions where participants create solutions to research problems — as an innovative teaching method. The project advances the TIRA platform with AI and Big Data capabilities, uniting research and teaching through problem-oriented, game-based learning.
TIMALIE — TIered MAchine Learning ArchItEctures
Developing tiered and continual learning methods — capsule networks, tree-based CNNs, and neural-network-based gradient boosting — that can learn new tasks without storing old data, enabling GDPR-compliant deployment. Applied to industrial quality assurance, autonomous robot navigation, and resource-constrained systems. Best Paper Award at Canadian AI 2022.
DIBS — Data MIning zur Beratung von Studierenden
Data mining approach to recognize dropout risks at early stages while prioritizing data protection.
SimCloud — Cloud-Based FEM Simulation
Making compute-intensive FEM simulations available as a cloud service while protecting sensitive engineering data. Security by design: domain decomposition distributes sub-problems across multiple cloud providers so that no single provider can reconstruct the original model. Machine learning optimizes load distribution across heterogeneous cloud resources.
