Research map / Architectures and efficiency / CNN design
VolterraNet: A Higher Order Convolutional Network With Group Equivariance for Homogeneous Manifolds
Monami Banerjee, Rudrasis Chakraborty, Jose Bouza, Baba C. Vemuri
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 VolterraNet instead of HCNN | P3 classifier · Synthetic Gaussians on P3 | mean accuracy: 86.50 -> 91.50 | Improved |
| use VolterraNet instead of Clebsch-Gordan net | spherical network · Spherical MNIST (NR/R) | accuracy: 95.86 -> 96.10 | Improved |
| use VolterraNet instead of Spherical CNN | spherical network · Spherical MNIST (R/R) | accuracy: 93.40 -> 96.71 | Improved |
| use VolterraNet instead of Clebsch-Gordan net on non-rotated data | spherical network · Spherical MNIST (NR/NR) | accuracy: 96.00 -> 96.72 | Hurt |
| use VolterraNet instead of Clebsch-Gordan net for parameter count | spherical network · Spherical MNIST | number of parameters: 342086 -> 46010 | Improved |
| use Euclidean baseline CNN on rotated test data | baseline CNN · Spherical MNIST | accuracy: 97.67 -> 22.18 | Hurt |
| add triplet loss to VolterraNet | VolterraNet · Shrec17 | P@N: 0.71 -> 0.73 | Improved |
| use VolterraNet with triplet loss instead of Esteves et al. | Shrec17 model · Shrec17 | P@N: 0.72 -> 0.73 | Improved |
| use VolterraNet instead of Spherical CNN on QM7 | atomic energy regression · QM7 | MSE: 8.47 -> 5.92 | Improved |
| make S2 layer second order on top of second-order SO(3) layer (ablation) | Dilated-VolterraNet · dMRI group test LM1 | p-value: 0.04 -> 0.01 | No clear effect |
| make SO(3) layer second order (ablation) | Dilated-VolterraNet · dMRI group test LM1 | p-value: 0.13 -> 0.01 | Improved |
| use dilated instead of standard VolterraNet (ablation) | VolterraNet · dMRI group test | parameters: p-values of 0.03 and 0.04 for LM1 and RM1 | Improved |
Closest papers by meaning
- Lie Group Decompositions for Equivariant Neural Networks (ICLR 2024 · similarity 0.87)
- Building Transformation Layers for Riemannian Neural Networks (NeurIPS 2026 · similarity 0.87)
- On the Fourier analysis in the SO(3) space : the EquiLoPO Network (ICLR 2025 · similarity 0.87)
- Variational Partial Group Convolutions for Input-Aware Partial Equivariance of Rotations and Color-Shifts (ICML 2024 · similarity 0.87)
- Equivariant Neural Tangent Kernels (ICML 2025 · similarity 0.86)
- Fast, Expressive $\mathrm{SE}(n)$ Equivariant Networks through Weight-Sharing in Position-Orientation Space (ICLR 2024 · similarity 0.86)
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.
