Research group on Human-Centric Machine Learning

The research group on Human-Centric Machine Learning at TUHH is dedicated to the study of the interaction between human users and AI systems. The main goal is to make AI systems usable by human practitioners who are not necessarily AI experts, by making the AI agents able to understand us better. This research is at the intersection between (multi-agent) reinforcement learning, cognitive modeling and human-computer interaction. 

Projects and theses

If you are interested in working on a research project or writing your thesis in the group, please have a look at the offered topics and book a meeting with me. 

News

11.09.25
Our paper "Detect, Adapt, Overcome: Mitigating Concept Drift in Federated Learning" got accepted for publication at the FLTA conference.
placeholder
11.09.25
Our paper "Single-positive Multi-label Learning with Label Cardinality" got accepted for publication at TMLR.
08.08.25
The abstract "Legal Co-pilots: Perspectives and Technical Challenges" was accepted for presentation at the Tübingen conference on AI and Law, which will take place in November.
14.04.25
Our paper "Two-Agent Case-Based Reasoning for Prediction" got accepted for presentation at the International Conference on Case-Based Reasoning (ICCBR).
04.04.25
We are glad to welcome Mahmoud Ajami, who is joining the group as a PhD student in collaboration with XFEL to explore the role of AI in experimental operations. He will be co-supervised with Steffen Hauf. Mahmoud's research will focus on the development of AI co-pilots ('whispering AIs') to support free-electron laser (FEL) experiments by identifying issues and suggesting actions. In the future, the developped tools could serve as a knowledge assistant and potentially steer parts of the experiment through human interaction.