Erik Bekkers

Erik Bekkers is an Associate Professor at the University of Amsterdam's Informatics Institute, leading research in the Machine Learning Lab (AMLab). His work bridges geometric mathematics and machine learning, focusing on developing robust and efficient deep learning architectures grounded in symmetry, equivariance, and physical principles.

• Education: PhD in Biomedical Engineering (cum laude) from Eindhoven University of Technology
• Previous Roles: Postdoctoral researcher in applied differential geometry at TU/e Department of Applied Mathematics

His research spans:

• Group convolutional neural networks
• Symmetry-preserving representation learning
• Generative modeling on manifolds
• Physics-informed neural networks
• Medical imaging applications

Recent publications emphasize geometric latent variable models, equivariant diffusion methods, and applications to molecular generation, medical imaging, and physics-driven AI. His team actively explores structure-preserving and self-supervised learning techniques.

Scientific Awards

• MICCAI Young Scientist Award (2018)
• Philips Impact Award (MIDL 2018)
• NWO VENI grant: Context-Aware AI in Medical Imaging (2023)
• NWO VIDI grant: Neural Ideograms - Geometry-Grounded AI (2024)

As co-founder of the ICML'24 GRaM workshop, he promotes geometry-grounded approaches in AI. His lab actively investigates geometric regularization, manifold-based PDE forecasting, and symmetry-aware generative methods.