Research map / Architectures and efficiency / CNN design
Clebsch–Gordan Nets: a Fully Fourier Space Spherical Convolutional Neural Network
Risi Kondor, Zhen Lin, Shubhendu Trivedi
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use fully Fourier-space Clebsch–Gordan layers instead of the Spherical CNN architecture | Spherical image classification · Rotated MNIST on the Sphere NR/NR | accuracy: 95.59 -> 96.4 | Improved |
| Use fully Fourier-space Clebsch–Gordan layers instead of the Spherical CNN architecture | Spherical image classification · Rotated MNIST on the Sphere NR/R | accuracy: 94.62 -> 96 | Improved |
| Use fully Fourier-space Clebsch–Gordan layers instead of the Spherical CNN architecture | Spherical image classification · Rotated MNIST on the Sphere R/R | accuracy: 93.4 -> 96.6 | Improved |
| Use fully Fourier-space Clebsch–Gordan layers instead of the Spherical CNN architecture | QM7 atomization energy prediction · QM7 | RMSE: 8.47 -> 7.97 | Improved |
| Use fully Fourier-space Clebsch–Gordan layers instead of the Spherical CNN architecture | QM7 atomization energy prediction · QM7 | RMSE: 5.96 -> 7.97 | Hurt |
| Use fully Fourier-space Clebsch–Gordan layers instead of task-specialized shape models | 3D shape recognition · SHREC17 | P@N: 0.814 -> 0.707 | Hurt |
| Use fully Fourier-space Clebsch–Gordan layers instead of task-specialized shape models | 3D shape recognition · SHREC17 | R@N: 0.769 -> 0.722 | Hurt |
| Use fully Fourier-space Clebsch–Gordan layers instead of task-specialized shape models | 3D shape recognition · SHREC17 | F1@N: 0.719 -> 0.701 | Hurt |
| Use fully Fourier-space Clebsch–Gordan layers instead of task-specialized shape models | 3D shape recognition · SHREC17 | mAP: 0.696 -> 0.683 | Hurt |
| Use fully Fourier-space Clebsch–Gordan layers instead of task-specialized shape models | 3D shape recognition · SHREC17 | NDCG: 0.783 -> 0.756 | Hurt |
Closest papers by meaning
- Efficient Generalized Spherical CNNs (ICLR 2021 · similarity 0.88)
- Spherical CNNs (ICLR 2018 · similarity 0.88)
- Spin-Weighted Spherical CNNs (NeurIPS 2020 · similarity 0.88)
- Scattering Networks on the Sphere for Scalable and Rotationally Equivariant Spherical CNNs (ICLR 2022 · similarity 0.88)
- Scalable and Equivariant Spherical CNNs by Discrete-Continuous (DISCO) Convolutions (ICLR 2023 · similarity 0.88)
- Rotation Equivariant Graph Convolutional Network for Spherical Image Classification (CVPR 2020 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
