Research map / Graph learning / GNN architectures
How Can State Space Models Enhance Machine Learning on Graphs?
Yinan Huang, Siqi Miao, Pan Li
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 |
|---|---|---|---|
| replace full-attention graph transformer module with GSSC (ablation) | GraphGPS · CIFAR10 | accuracy: 72.298 -> 77.642 | Improved |
| replace attention with GSSC (ablation) | GraphGPS · PascalVOC-SP | F1: 0.3748 -> 0.4561 | Improved |
| replace attention with GSSC (ablation) | GraphGPS · Peptides-func | AP: 0.6535 -> 0.7081 | Improved |
| replace attention with GSSC (ablation) | GraphGPS · ZINC-12k | MAE: 0.070 -> 0.064 | Improved |
| replace attention with GSSC (ablation) | GraphGPS · ogbg-molhiv | AUROC: 78.80 -> 80.35 | Improved |
| add GSSC to an MPNN to predict 5-cycles | GIN · Cycle counting 5-cycle | normalized MAE: 0.2088 -> 0.0113 | Improved |
| add GSSC compared with ID-GNN on 5-cycles | ID-GIN · Cycle counting 5-cycle | normalized MAE: 0.0490 -> 0.0113 | Improved |
| GSSC versus I2-GNN on 5-cycle counting | I2-GNN · Cycle counting 5-cycle | normalized MAE: 0.0028 -> 0.0113 | Hurt |
| GSSC versus I2-GNN on 3-cycle counting | I2-GNN · Cycle counting 3-cycle | normalized MAE: 0.0003 -> 0.0002 | Improved |
| GSSC versus Grit on CLUSTER | Grit · CLUSTER | accuracy: 80.026 -> 79.156 | Hurt |
| GSSC versus GRED on Peptides-func | GRED · Peptides-func | AP: 0.7133 -> 0.7081 | Hurt |
| GSSC versus Graph-Mamba-I on CIFAR10 | Graph-Mamba-I · CIFAR10 | accuracy: 73.700 -> 77.642 | Improved |
Closest papers by meaning
- State Space Models on Temporal Graphs: A First-Principles Study (NeurIPS 2024 · similarity 0.91)
- Spatio-Spectral Graph Neural Networks (NeurIPS 2024 · similarity 0.88)
- Recurrent Distance Filtering for Graph Representation Learning (ICML 2024 · similarity 0.87)
- Two Heads Are Better Than One: Boosting Graph Sparse Training via Semantic and Topological Awareness (ICML 2024 · similarity 0.87)
- Generative Graph Pattern Machine (NeurIPS 2025 · similarity 0.87)
- GG-SSMs: Graph-Generating State Space Models (CVPR 2025 · similarity 0.87)
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.
