Dr. Wendelin Böhmer

Biography

Wendelin Böhmer studied computer science and received his PhD from the Technical University of Berlin. He worked as a postdoctoral researcher at the University of Oxford before joining the Delft University of Technology as an Assistant Professor in the Algorithmics group within the Department of Software Technology. His research interests lie at the intersection of inductive and deductive reasoning in Artificial Intelligence (AI). Traditionally, deductive approaches like Operations Research (OR) have dominated AI, but in recent decades, inductive approaches such as Machine Learning (ML) have gained public attention. New techniques are addressing fundamental flaws in deductive reasoning, particularly in how they align with reality and their inherent limitations. One of the most visible aspects of this shift is Reinforcement Learning (RL), which learns from interactions within potentially non-deterministic environments. Böhmer's research focuses on tackling the challenges of inductive reasoning, especially in the context of RL, which presents unique issues such as generalization from limited examples and adapting to changing environments. He is particularly interested in developing 'imaginative' agents that employ deductive reasoning based on inner abstractions and consistently testing their models against reality. His current work emphasizes the use of Graph Neural Networks, Attention Architectures, and Ensemble Estimates to enhance the performance and applicability of these agents in fields like robotics. Böhmer is eager to collaborate with others interested in these areas.

Research Interests