Equivariant Geometric Scattering Networks via Vector Diffusion Wavelets
This work addresses the need for efficient and symmetric neural networks for geometric graphs in fields like molecular modeling, though it appears incremental as it builds on existing scattering and GNN methods.
The authors tackled the problem of designing a geometric scattering transform for graphs with scalar and vector features that is equivariant to rigid-body transformations, and they showed that their equivariant scattering-based GNN achieves performance comparable to other equivariant GNNs while using significantly fewer parameters.
We introduce a novel version of the geometric scattering transform for geometric graphs containing scalar and vector node features. This new scattering transform has desirable symmetries with respect to rigid-body roto-translations (i.e., $SE(3)$-equivariance) and may be incorporated into a geometric GNN framework. We empirically show that our equivariant scattering-based GNN achieves comparable performance to other equivariant message-passing-based GNNs at a fraction of the parameter count.