Research map / Graph learning / Graph self-supervised learning
Structure-Aware Transformer for Graph Representation Learning
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 |
|---|---|---|---|
| add structure-aware attention using k-subgraph representations | PNA Transformer on ZINC · ZINC | MAE: 0.188$\pm$0.004 -> 0.094$\pm$0.008 | Helped |
| add structure-aware attention using k-subgraph representations | GCN on CLUSTER · CLUSTER | Accuracy: 68.498$\pm$0.976 -> 77.682$\pm$0.098 | Helped |
| add structure-aware attention using k-subgraph representations | GCN on PATTERN · PATTERN | Accuracy: 71.892$\pm$0.334 -> 86.816$\pm$0.028 | Helped |
| add structure-aware attention using k-subtree representations | GCN on OGBG-PPA · OGBG-PPA | Accuracy: 0.6839$\pm$0.0084 -> 0.7483$\pm$0.0048 | Helped |
| add structure-aware attention using k-subtree representations | GCN on OGBG-CODE2 · OGBG-CODE2 | F1 score: 0.1507$\pm$0.0018 -> 0.1934$\pm$0.0020 | Helped |
| increase subgraph radius from k=0 to a positive k (ablation) | SAT with PNA on ZINC · ZINC | MAE | Helped |
| increase k beyond k=4 in the k-subtree extractor (ablation) | SAT with PNA on ZINC · ZINC | MAE | Hurt |
| add absolute positional encoding (ablation) | SAT with PNA on ZINC · ZINC | MAE | Helped |
| replace mean readout with sum or CLS readout (ablation) | SAT with PNA on ZINC · ZINC | MAE: 0.131$\pm$0.002 -> 0.129$\pm$0.003 (sum); 0.131$\pm$0.002 -> 0.128$\pm$0.004 (CLS) | No clear effect |
Closest papers by meaning
- What Improves the Generalization of Graph Transformers? A Theoretical Dive into the Self-attention and Positional Encoding (ICML 2024 · similarity 0.92)
- Transitivity-Preserving Graph Representation Learning for Bridging Local Connectivity and Role-Based Similarity (AAAI 2024 · similarity 0.91)
- Towards Principled Graph Transformers (NeurIPS 2024 · similarity 0.91)
- Subgraphormer: Unifying Subgraph GNNs and Graph Transformers via Graph Products (ICML 2024 · similarity 0.90)
- Generalizable Insights for Graph Transformers in Theory and Practice (NeurIPS 2025 · similarity 0.90)
- Translating Subgraphs to Nodes Makes Simple GNNs Strong and Efficient for Subgraph Representation Learning (ICML 2024 · similarity 0.89)
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.