Erik J. Bekkers
Papers
1
Total Citations
4
H-Index
1
About
Erik J. Bekkers is a leading researcher at the intersection of geometric deep learning, mathematical imaging, and PDE-based modeling. His work centers on developing principled mathematical frameworks for processing data on non-Euclidean spaces, particularly the homogeneous space of 3D positions and orientations. Bekkers’ major contribution lies in deriving exact analytic solutions to Fokker-Planck PDEs on this joint space, enabling stable Lévy and Wiener processes that underpin advances in robotics, mechanics, and image analysis. His 2018 paper on Fourier transforms for these spaces, despite its recent publication, has already garnered 4 citations, reflecting its foundational impact. Notably, Bekkers is also recognized for pioneering the concept of "group equivariant convolutional networks," bridging group theory and deep learning to create models that inherently respect symmetries in data. His work has been instrumental in advancing directional statistics and probabilistic modeling, offering both theoretical elegance and practical tools for complex spatial reasoning. For students and researchers, Bekkers exemplifies how deep mathematical insight can drive innovation in applied AI and computational science.
Research Focus
Key Achievements
Top Papers
- 1