ConceptGeometric Deep Learning
Geometric Deep Learning
A field that extends deep learning to non-Euclidean data such as graphs, manifolds and point clouds, emphasizing the use of symmetry and equivariance as inductive biases.
Timeline (last 90 days)
Sep 2
Geometric deep learning was systematically proposed by scholars including Michael Bronstein and Joan Bruna, with the core idea that deep learning architectures can be understood as encoding equivariance to specific symmetry groups.
Unverified50%