Research map / Graph learning / GNN architectures
SE(3) Equivariant Graph Neural Networks with Complete Local Frames
Weitao Du, He Zhang, Yuanqi Du, Qi Meng, Wei Chen, Nanning Zheng, Bin Shao, Tie-Yan Liu
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 complete local frames instead of radial-only geometric features | ClofNet · ES(20) | MSE: 0.0128 -> 0.0073 | Improved |
| use complete local frames instead of radial-only geometric features | ClofNet · G+ES(20) | MSE: 0.0118 -> 0.0072 | Improved |
| use complete local frames instead of radial-only geometric features | ClofNet · L+ES(20) | MSE: 0.0368 -> 0.0251 | Improved |
| use complete local frames instead of radial-only geometric features | ClofNet · ES(5) | MSE: 0.0079 -> 0.0065 | Improved |
| use ClofNet rather than DGSM | molecular conformation generation · GEOM-QM9 | COV-mean: 91.49 -> 90.21 | Hurt |
| use ClofNet rather than DGSM | molecular conformation generation · GEOM-QM9 | MAT-mean: 0.2139 -> 0.2430 | Hurt |
| use ClofNet rather than DGSM | molecular conformation generation · GEOM-Drugs | COV-mean: 78.73 -> 88.64 | Improved |
| use ClofNet rather than DGSM | molecular conformation generation · GEOM-Drugs | MAT-mean: 1.0154 -> 0.9040 | Improved |
| use ClofNet rather than ConfGF | distance distribution modeling · ISO17 | MMD-Single mean: 0.3684 -> 0.1317 | Improved |
| add scalarization block (ablation) | ClofNet · GEOM-QM9 | COV-mean: 88.99 -> 90.21 | Improved |
| add scalarization block (ablation) | ClofNet · GEOM-Drugs | COV-mean: 65.67 -> 88.64 | Improved |
| add graph transformer attention instead of using GIN (ablation) | ClofNet · GEOM-QM9 | COV-mean: 85.21 -> 90.21 | Improved |
Closest papers by meaning
- Are High-Degree Representations Really Unnecessary in Equivariant Graph Neural Networks? (NeurIPS 2024 · similarity 0.91)
- Universally Invariant Learning in Equivariant GNNs (NeurIPS 2025 · similarity 0.91)
- On the Generalization of Equivariant Graph Neural Networks (ICML 2024 · similarity 0.89)
- A Characterization Theorem for Equivariant Networks with Point-wise Activations (ICLR 2024 · similarity 0.89)
- Permutation Equivariant Neural Networks for Symmetric Tensors (ICML 2025 · similarity 0.88)
- SEGNO: Generalizing Equivariant Graph Neural Networks with Physical Inductive Biases (ICLR 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.
