Learning SO(3) Equivariant Representations with Spherical CNNs

Keyphrases

• Data Augmentation 100%
• SO(3) 100%
• 3D Rotation 100%
• Equivariant Representation 100%
• Spherical CNN 100%
• Multi-valued 50%
• High Capacity 50%
• Classification Task 50%
• Convolutional Neural Network 50%
• Local Symmetry 50%
• 3D Classification 50%
• Spectral Domain 50%
• Spherical Functions 50%
• 3D Data 50%
• Low Capacity 50%
• Nuisance 50%
• Classification Benchmark 50%
• Retrieval Benchmark 50%
• 3D Shape Retrieval 50%
• 3D Shape Classification 50%
• Spherical Harmonic Domain 50%
• Smooth Spectrum 50%
• Spherical Convolutional Networks 50%
• Rotational Equivariance 50%

Computer Science

• Data Augmentation 100%
• Convolutional Neural Network 100%
• Optical Spectrum 50%
• Convolutional Network 50%
• Classification Task 50%
• Spectral Domain 50%
• Spherical Harmonic 50%
• Shape Retrieval 50%
• Resulting Filter 50%
• Convolution 50%

Mathematics

• Equivariant 100%
• Convolutional Neural Network 100%
• Data Augmentation 100%
• Spherical Harmonic 50%
• Convolution 50%
• Spherical Function 50%
• 3D Shape 50%