Research map / Graph learning / GNN architectures
E(n) Equivariant Graph 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 |
|---|---|---|---|
| Use EGNN instead of Radial Field | N-body position forecasting · N-body system | MSE: 0.0104 -> 0.0071 | Helped |
| Use EGNN instead of a non-equivariant GNN | N-body position forecasting · N-body system | MSE: 0.0107 -> 0.0071 | Helped |
| Use EGNN instead of DimeNet++ | QM9 property prediction · QM9 | μ MAE: .030 -> .029 | Helped |
| Use EGNN instead of DimeNet++ | QM9 property prediction · QM9 | R² MAE: .331 -> .106 | Hurt |
| Use EGNN instead of DimeNet++ | QM9 property prediction · QM9 | α MAE: .044 -> .071 | Hurt |
| Use EGNN instead of Noise-GNN | Graph autoencoding · Community Small, n=8 | % Err.: 0.44 -> 0.06 | Helped |
| Use EGNN instead of Noise-GNN | Graph autoencoding · Erdos&Renyi, n=8 | % Err.: 1.25 -> 0.11 | Helped |
| Add input noise to the GNN (ablation) | Graph autoencoding · Erdos&Renyi | % Err.: 7.92 -> 3.80 | Helped |
| Increase embedding dimension from n=4 to n=8 (ablation) | EGNN graph autoencoder · Erdos&Renyi | % Err.: 3.09 -> 0.11 | Helped |
| Increase embedding dimension from n=4 to n=8 (ablation) | EGNN graph autoencoder · Community Small | % Err.: 2.19 -> 0.06 | Helped |
| Update coordinates in the QM9 model rather than keeping positions static (ablation) | EGNN molecular property predictor · QM9 | MAE | No clear effect |
| Use equivariant EGNN rather than GNN when fitting graph autoencoding training data | Graph autoencoder · Erdos&Renyi, 100-graph training subset | F1 | Helped |
Closest papers by meaning
- Are High-Degree Representations Really Unnecessary in Equivariant Graph Neural Networks? (NeurIPS 2024 · similarity 0.92)
- On the Generalization of Equivariant Graph Neural Networks (ICML 2024 · similarity 0.92)
- Universally Invariant Learning in Equivariant GNNs (NeurIPS 2025 · similarity 0.91)
- A Characterization Theorem for Equivariant Networks with Point-wise Activations (ICLR 2024 · similarity 0.90)
- Permutation Equivariant Neural Controlled Differential Equations for Dynamic Graph Representation Learning (NeurIPS 2025 · similarity 0.88)
- Schur Nets: exploiting local structure for equivariance in higher order graph neural networks (NeurIPS 2024 · similarity 0.88)
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.