Research map / Graph learning / GNN architectures
Proper Velocity Neural Networks
Ziheng Chen, Zihan Su, Bernhard Schölkopf, Nicu Sebe
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 Riemannian PV FC layer instead of tangent-space FC (ablation) | PVNN · Airport node classification | accuracy (%): 86.99 -> 97.93 | Improved |
| use Riemannian PV FC layer instead of tangent-space FC (ablation) | PVNN · Cora node classification | accuracy (%): 53.58 -> 52.26 | Hurt |
| use GyroBN instead of tangent batch norm (ablation) | PVNN · Airport node classification | accuracy (%): 98.71 -> 99.03 | Improved |
| use GyroBN instead of tangent batch norm (ablation) | PVNN · Cora node classification | accuracy (%): 45.36 -> 46.64 | Improved |
| lift input features with Exp0 before PVNN (ablation) | PVNN · Airport node classification | accuracy (%): 97.71 -> 97.93 | No clear effect |
| lift backbone output with Exp0 in the PV MLR head (ablation) | PV MLR · CIFAR-100 | accuracy (%): 78.20 -> 78.19 | No clear effect |
| use tangent activation instead of Euclidean activation in PV (ablation) | PVNN · Cora node classification | accuracy (%): 38.10 -> 52.26 | Improved |
| add tangent activation on top of the FC internal nonlinearity (ablation) | PVNN · Airport node classification | accuracy (%): 99.40 -> 99.15 | No clear effect |
| use Frechet mean with 10 iterations instead of Euclidean batch statistics (ablation) | PV GyroBN · PubMed node classification | accuracy (%): 69.82 -> 74.34 | Improved |
| use Frechet mean with 10 iterations instead of Euclidean batch statistics (cost) (ablation) | PV GyroBN · Airport fit time | ms per epoch: 55.19 -> 105.79 | Hurt |
| increase Frechet iterations from 1 to 10 (ablation) | PV GyroBN · Airport node classification | accuracy (%): 88.93 -> 99.03 | Improved |
| run Frechet solver to convergence instead of 10 iterations (ablation) | PV GyroBN · Airport node classification | accuracy (%): 99.03 -> 98.46 | No clear effect |
Closest papers by meaning
- Riemann Tensor Neural Networks: Learning Conservative Systems with Physics-Constrained Networks (ICML 2025 · similarity 0.87)
- A Riemannian Framework for Learning Reduced-order Lagrangian Dynamics (ICLR 2025 · similarity 0.85)
- Matrix Manifold Neural Networks++ (ICLR 2024 · similarity 0.85)
- Clifford-Steerable Convolutional Neural Networks (ICML 2024 · similarity 0.85)
- Riemannian Multinomial Logistics Regression for SPD Neural Networks (CVPR 2024 · similarity 0.85)
- RMLR: Extending Multinomial Logistic Regression into General Geometries (NeurIPS 2024 · similarity 0.85)
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.
