معرفی
Dr. Robert Vandermeulen is a Postdoctoral Researcher at the Berlin Institute for the Foundations of Learning and Data (BIFOLD), Technical University of Berlin. His research focuses on anomaly detection, nonparametric statistics, and aligning neural networks with human cognitive processes. He holds a PhD and two Master's degrees from the University of Michigan in Electrical Engineering and Mathematics.
Education:
- PhD in Electrical Engineering, University of Michigan (2016)
- MS in Mathematics, University of Michigan (2015)
- MS in Electrical Engineering, University of Michigan (2012)
Research Interests:
- Deep anomaly detection in high-dimensional data
- Nonparametric density estimation and tensor methods
- Human vs. neural network alignment for interpretable AI
- Applications in medical imaging and explainable AI
Key Research Trends: His recent work addresses challenges in anomaly detection across domains like healthcare imaging and text analysis. He explores nonparametric methods to overcome the curse of dimensionality and integrates human cognitive insights to improve neural network robustness. Contributions include novel frameworks like input Hessian regularization and VICE (Variational Inference for Concept Embeddings).
Affiliations & Prior Work: Previously a postdoctoral researcher at Technische Universität Kaiserslautern, he collaborates with BIFOLD on foundational machine learning research. His work bridges theoretical statistics and applied deep learning, with a focus on interpretable and reliable AI systems.
Labs & Teams: Affiliated with BIFOLD, a cross-disciplinary institute advancing learning theory and data science fundamentals at TU Berlin.
