Research map / Graph learning / GNN architectures
Equivariant Graph Hierarchy-Based Neural Networks
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| remove hierarchy (w/o Hierarchy) (ablation) | EGHN · Mocap Walk | MSE: 21.9 -> 8.5 | Helped |
| remove hierarchy (w/o Hierarchy) (ablation) | EGHN · Mocap Run | MSE: 42.1 -> 25.9 | Helped |
| remove equivariance (ablation) | EGHN · Mocap Walk | MSE: 19.7 -> 8.5 | Helped |
| remove connectivity loss (ablation) | EGHN · Mocap Walk | MSE: 10.5 -> 8.5 | Helped |
| use only global adjacency instead of both (ablation) | EGHN · Mocap Walk | MSE: 8.5 -> 17.4 | Hurt |
| use only local adjacency instead of both (ablation) | EGHN · Mocap Walk | MSE: 8.5 -> 16.8 | Hurt |
| replace EGHN by EGNN (ablation) | EGHN · Mocap Walk | MSE: 22.3 -> 8.5 | Helped |
| number of clusters K=3 vs 5 (ablation) | EGHN · Mocap Walk | MSE: 10.1 -> 8.5 | Helped |
| number of clusters K=8 vs 5 (ablation) | EGHN · Mocap Walk | MSE: 14.9 -> 8.5 | Helped |
| sweep clusters 1 to 5 on Complex (3, 3) (ablation) | EGHN · Simulation Complex (3, 3) | MSE: 14.86 -> 11.58 | Mixed |
| sweep clusters K on Protein MD (ablation) | EGHN · Protein MD | MSE: 2.132 -> 1.843 | Mixed |
| hierarchy via cluster assignment vs identity (w/o hier) (ablation) | EGHN · Complex (3,3) | MSE: 12.24 -> 11.58 | Helped |
Closest papers by meaning
- SEGNO: Generalizing Equivariant Graph Neural Networks with Physical Inductive Biases (ICLR 2024 · similarity 0.89)
- Universally Invariant Learning in Equivariant GNNs (NeurIPS 2025 · similarity 0.89)
- Are High-Degree Representations Really Unnecessary in Equivariant Graph Neural Networks? (NeurIPS 2024 · similarity 0.89)
- Equivariant Graph Neural Operator for Modeling 3D Dynamics (ICML 2024 · similarity 0.88)
- Improving Equivariant Graph Neural Networks on Large Geometric Graphs via Virtual Nodes Learning (ICML 2024 · similarity 0.87)
- Long-Short-Range Message-Passing: A Physics-Informed Framework to Capture Non-Local Interaction for Scalable Molecular Dynamics Simulation (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.