Equivariant Maps for Hierarchical Structures
arXiv:2006.03627
Abstract
While using invariant and equivariant maps, it is possible to apply deep learning to a range of primitive data structures, a formalism for dealing with hierarchy is lacking. This is a significant issue because many practical structures are hierarchies of simple building blocks; some examples include sequences of sets, graphs of graphs, or multiresolution images. Observing that the symmetry of a hierarchical structure is the "wreath product" of symmetries of the building blocks, we express the equivariant map for the hierarchy using an intuitive combination of the equivariant linear layers of the building blocks. More generally, we show that any equivariant map for the hierarchy has this form. To demonstrate the effectiveness of this approach to model design, we consider its application in the semantic segmentation of point-cloud data. By voxelizing the point cloud, we impose a hierarchy of translation and permutation symmetries on the data and report state-of-the-art on Semantic3D, S3DIS, and vKITTI, that include some of the largest real-world point-cloud benchmarks.
References in corpus (13)
- Tensor field networks: Rotation- and translation-equivariant neural networks for 3D point clouds
- How Powerful are Graph Neural Networks?
- Submanifold Sparse Convolutional Networks
- Virtual Worlds as Proxy for Multi-Object Tracking Analysis
- On the Generalization of Equivariance and Convolution in Neural Networks to the Action of Compact Groups
- Augmented LiDAR Simulator for Autonomous Driving
- General -Equivariant Steerable CNNs
- Covariant Compositional Networks For Learning Graphs
- Deep Models of Interactions Across Sets
- Janossy Pooling: Learning Deep Permutation-Invariant Functions for Variable-Size Inputs
- On Learning Sets of Symmetric Elements
- Deep Learning for 3D Point Clouds: A Survey
- Equivariant Entity-Relationship Networks