Research map / Graph learning / GNN architectures
VDW-GNNs: Vector diffusion wavelets for geometric graph neural networks
David R Johnson, Alexander Sietsema, Rishabh Anand, Deanna Needell, Smita Krishnaswamy, Michael Perlmutter
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 |
|---|---|---|---|
| add rotational equivariance (vector-track wavelets) instead of treating components as scalars (ablation) | VDW-GNN · Ellipsoid diameter (graph-level) | rotated test MSE: 2.9181 -> 0.0037 | Improved |
| add rotational equivariance instead of treating components as scalars (ablation) | VDW-GNN · Ellipsoid normal-field (node-level) | rotated test MSE: 0.3552 -> 0.1513 | Improved |
| use VDW-GNN instead of EGNN with far fewer parameters | VDW-GNN · Ellipsoid diameter (graph-level) | parameter count: 59,779 vs 284,036 | Improved |
| use VDW-GNN instead of TFN on node-level task | VDW-GNN · Ellipsoid normal-field (node-level) | rotated test MSE: 0.1513 vs 0.1145 | Hurt |
| use VDW-GNN instead of DD-TNN on wind reconstruction | VDW-GNN · Wind field reconstruction | test MSE: 12.7592 -> 2.7546 | Improved |
| use VDW-GNN instead of MARBLE on neural decoding | VDW-GNN · Macaque neural recordings (44 days, 3-D embedding) | classifier accuracy: 0.66 vs 0.65 | No clear effect |
| use VDW-GNN instead of EGNN on wind reconstruction (rotated test) | VDW-GNN · Wind field reconstruction | rotated test MSE: 3.2266 vs 2.7651 | Hurt |
Closest papers by meaning
- Hyperbolic-PDE GNN: Spectral Graph Neural Networks in the Perspective of A System of Hyperbolic Partial Differential Equations (ICML 2025 · similarity 0.87)
- Hyperbolic Geometric Latent Diffusion Model for Graph Generation (ICML 2024 · similarity 0.86)
- SEA-GWNN: Simple and Effective Adaptive Graph Wavelet Neural Network (AAAI 2024 · similarity 0.86)
- WaveNet: Tackling Non-stationary Graph Signals via Graph Spectral Wavelets (AAAI 2024 · similarity 0.85)
- Toward a Unified Geometry Understanding : Riemannian Diffusion Framework for Graph Generation and Prediction (NeurIPS 2025 · similarity 0.85)
- Feature Transportation Improves Graph Neural Networks (AAAI 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.
