Mario Geiger

Papers

Total Citations

H-Index

About

Mario Geiger is a leading researcher in geometric deep learning, best known for pioneering work on neural networks that respect the symmetries of non-Euclidean domains. His most influential contribution is the development of Spherical CNNs (2018), which introduced a principled framework for applying convolutional neural networks to spherical signals—a critical advance for omnidirectional vision in drones, robotics, and autonomous systems. This paper has garnered 38 citations and laid the foundation for subsequent work on rotation-equivariant architectures. Geiger’s research focuses on building machine learning models that exploit the underlying symmetries of data, particularly through group-equivariant and steerable neural networks. His work has enabled more sample-efficient and robust learning on spheres, graphs, and 3D shapes, with applications spanning computer vision, physics, and molecular modeling. By combining rigorous mathematical theory with practical algorithmic design, Geiger has helped shape the emerging field of geometric deep learning, making him a key figure for students and researchers interested in symmetry-aware AI.