Gauge Equivariant Mesh CNNs: Anisotropic convolutions on geometric graphs

Abstract

A common approach to define convolutions on meshes is to interpret them as a graph and apply graph convolutional networks (GCNs). Such GCNs utilize isotropic kernels and are therefore insensitive to the relative orientation of vertices and thus to the geometry of the mesh as a whole. We propose Gauge Equivariant Mesh CNNs which generalize GCNs to apply anisotropic gauge equivariant kernels. Since the resulting features carry orientation information, we introduce a geometric message passing scheme defined by parallel transporting features over mesh edges. Our experiments validate the significantly improved expressivity of the proposed model over conventional GCNs and other methods.

arXiv commentPublished at ICLR 2021

References in corpus (15)

• Geometric deep learning: going beyond Euclidean data
• Group Equivariant Convolutional Networks
• Tensor field networks: Rotation- and translation-equivariant neural networks for 3D point clouds
• 3D Steerable CNNs: Learning Rotationally Equivariant Features in Volumetric Data
• DeepSphere: Efficient spherical Convolutional Neural Network with HEALPix sampling for cosmological applications
• Gauge Equivariant Convolutional Networks and the Icosahedral CNN
• General $E(2)$-Equivariant Steerable CNNs
• 3D G-CNNs for Pulmonary Nodule Detection
• Deep Scale-spaces: Equivariance Over Scale
• Neural 3D Morphable Models: Spiral Convolutional Networks for 3D Shape Representation Learning and Generation
• Geometric deep learning on graphs and manifolds using mixture model CNNs
• A Wigner-Eckart Theorem for Group Equivariant Convolution Kernels
• Parallel Transport Convolution: A New Tool for Convolutional Neural Networks on Manifolds
• Coordinate Independent Convolutional Networks -- Isometry and Gauge Equivariant Convolutions on Riemannian Manifolds
• Steerable Partial Differential Operators for Equivariant Neural Networks

Cited by in corpus (9)

• Geometric Deep Learning: Grids, Groups, Graphs, Geodesics, and Gauges
• Improving Graph Neural Network Expressivity via Subgraph Isomorphism Counting
• Multipole Graph Neural Operator for Parametric Partial Differential Equations
• A Wigner-Eckart Theorem for Group Equivariant Convolution Kernels
• Large-scale graph representation learning with very deep GNNs and self-supervision
• Probabilistic Numeric Convolutional Neural Networks
• HodgeNet: Learning Spectral Geometry on Triangle Meshes
• Rotation-Invariant Autoencoders for Signals on Spheres
• Surrogate Modelling for Injection Molding Processes using Machine Learning