Equivariant Geometric Scattering Networks via Vector Diffusion Wavelets
Dave Johnson · Rishabh Anand · Smita Krishnaswamy · Michael Perlmutter
Abstract
We introduce a novel version of the geometric scattering transform for geometric graphs with both scalar- and vector-valued node features. We show that this new scattering transform has desirable symmetries with respect to rotations in Euclidean space and may be incorporated into a geometric GNN, which achieves comparable performance to other equivariant GNNs at a fraction of the parameter count.
Chat is not available.
Successful Page Load