3D-Rotation-Equivariant Quaternion Neural Networks
arXiv:1911.09040
Abstract
This paper proposes a set of rules to revise various neural networks for 3D point cloud processing to rotation-equivariant quaternion neural networks (REQNNs). We find that when a neural network uses quaternion features under certain conditions, the network feature naturally has the rotation-equivariance property. Rotation equivariance means that applying a specific rotation transformation to the input point cloud is equivalent to applying the same rotation transformation to all intermediate-layer quaternion features. Besides, the REQNN also ensures that the intermediate-layer features are invariant to the permutation of input points. Compared with the original neural network, the REQNN exhibits higher rotation robustness.
References in corpus (11)
- Tensor field networks: Rotation- and translation-equivariant neural networks for 3D point clouds
- PointSIFT: A SIFT-like Network Module for 3D Point Cloud Semantic Segmentation
- Dynamic Graph CNN for Learning on Point Clouds
- On the Generalization of Equivariance and Convolution in Neural Networks to the Action of Compact Groups
- Deep Complex Networks
- Spherical CNNs
- Full-Capacity Unitary Recurrent Neural Networks
- On Complex Valued Convolutional Neural Networks
- Complex Gated Recurrent Neural Networks
- Pointwise Rotation-Invariant Network with Adaptive Sampling and 3D Spherical Voxel Convolution
- Interpretable Complex-Valued Neural Networks for Privacy Protection
Cited by in corpus (7)
- A Rotation-Invariant Framework for Deep Point Cloud Analysis
- You Only Hypothesize Once: Point Cloud Registration with Rotation-equivariant Descriptors
- Coordinate Independent Convolutional Networks -- Isometry and Gauge Equivariant Convolutions on Riemannian Manifolds
- Interpretable Complex-Valued Neural Networks for Privacy Protection
- Rotation Transformation Network: Learning View-Invariant Point Cloud for Classification and Segmentation
- Rotation-Invariant Gait Identification with Quaternion Convolutional Neural Networks
- Verifiability and Predictability: Interpreting Utilities of Network Architectures for Point Cloud Processing