Gauge Equivariant Convolutional Networks and the Icosahedral CNN

Abstract

The principle of equivariance to symmetry transformations enables a theoretically grounded approach to neural network architecture design. Equivariant networks have shown excellent performance and data efficiency on vision and medical imaging problems that exhibit symmetries. Here we show how this principle can be extended beyond global symmetries to local gauge transformations. This enables the development of a very general class of convolutional neural networks on manifolds that depend only on the intrinsic geometry, and which includes many popular methods from equivariant and geometric deep learning. We implement gauge equivariant CNNs for signals defined on the surface of the icosahedron, which provides a reasonable approximation of the sphere. By choosing to work with this very regular manifold, we are able to implement the gauge equivariant convolution using a single conv2d call, making it a highly scalable and practical alternative to Spherical CNNs. Using this method, we demonstrate substantial improvements over previous methods on the task of segmenting omnidirectional images and global climate patterns.

Proceedings of the International Conference on Machine Learning (ICML), 2019

References in corpus (1)

• Spherical CNNs on Unstructured Grids

Cited by in corpus (36)

• A Survey of Sound Source Localization with Deep Learning Methods
• Stand-Alone Self-Attention in Vision Models
• Sampling using $SU(N)$ gauge equivariant flows
• Three-dimensional microstructure generation using generative adversarial neural networks in the context of continuum micromechanics
• CNNs on Surfaces using Rotation-Equivariant Features
• Efficient Modelling of Trivializing Maps for Lattice ${\varphi }^4$ Theory Using Normalizing Flows: A First Look at Scalability
• Scale-Equivariant Steerable Networks
• Permutationless Many-Jet Event Reconstruction with Symmetry Preserving Attention Networks
• DeltaConv: Anisotropic Operators for Geometric Deep Learning on Point Clouds
• Direction of Arrival Estimation of Sound Sources Using Icosahedral CNNs
• Data-driven emergence of convolutional structure in neural networks
• InMoDeGAN: Interpretable Motion Decomposition Generative Adversarial Network for Video Generation
• Co-Attentive Equivariant Neural Networks: Focusing Equivariance On Transformations Co-Occurring In Data
• Introduction to Normalizing Flows for Lattice Field Theory
• Rotation-Equivariant Deep Learning for Diffusion MRI
• Covariance in Physics and Convolutional Neural Networks
• Machine learning and invariant theory
• Category Theory in Machine Learning
• Equivariance and generalization in neural networks
• Topos and Stacks of Deep Neural Networks
• Group Equivariant Stand-Alone Self-Attention For Vision
• Affine Self Convolution
• ManifoldNorm: Extending normalizations on Riemannian Manifolds
• Rotation-Invariant Autoencoders for Signals on Spheres
• Equivariant Networks for Pixelized Spheres
• Preserving gauge invariance in neural networks
• RISA-Net: Rotation-Invariant Structure-Aware Network for Fine-Grained 3D Shape Retrieval
• Applying VertexShuffle Toward 360-Degree Video Super-Resolution on Focused-Icosahedral-Mesh
• The Physics of Machine Learning: An Intuitive Introduction for the Physical Scientist
• Efficient Generalized Spherical CNNs
• Equivariant Point Network for 3D Point Cloud Analysis
• ChebLieNet: Invariant Spectral Graph NNs Turned Equivariant by Riemannian Geometry on Lie Groups
• Computational Analysis of Deformable Manifolds: from Geometric Modelling to Deep Learning
• Surrogate Modelling for Injection Molding Processes using Machine Learning
• Computing equivariant matrices on homogeneous spaces for Geometric Deep Learning and Automorphic Lie Algebras
• Interpretable Image Clustering via Diffeomorphism-Aware K-Means