Research map / Graph learning / GNN architectures
Representing Long-Range Context for Graph Neural Networks with Global Attention
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 Transformer module after GNN (same GCN backbone) | GCN 3-layer · NCI1 | accuracy: 7.1% | Helped |
| add Transformer module after GIN-Virtual | GIN-Virtual · ogbg-molpcba test | AP: 0.2703 -> 0.2761 | Helped |
| add Transformer module after GCN-Virtual | GCN-Virtual · ogbg-code2 test | F1: 0.1595 -> 0.1830 | Helped |
| add Transformer module after PNA | PNA · ogbg-code2 test | F1: 0.1570 -> 0.1765 | Helped |
| use Transformer alone without GNN (ablation) | GraphTrans · ogbg-molpcba test | AP: 0.2756 -> 0.1281 | Hurt |
| add Transformer on top of frozen pretrained GNN (ablation) | GCN-Virtual · ogbg-code2 test | F1: 0.1574 -> 0.1616 | Helped |
| fine-tune GNN jointly instead of freezing (ablation) | GraphTrans · ogbg-code2 test | F1: 0.1616 -> 0.1733 | Helped |
| use <CLS> readout instead of mean pooling (ablation) | GraphTrans · ogbg-code2 test | F1: 0.1509 -> 0.1784 | Helped |
| use <CLS> readout instead of last-token (ablation) | GraphTrans · ogbg-code2 test | F1: 0.1716 -> 0.1784 | Helped |
| concatenate input graph embeddings to Transformer input (ablation) | GraphTrans <CLS> · ogbg-code2 test | F1: 0.1784 -> 0.1810 | Helped |
| add Transformer module | GCN-Virtual · Code2 | training speed: 1.00x -> 0.93x | Hurt |
Closest papers by meaning
- Learning Long Range Dependencies on Graphs via Random Walks (ICLR 2025 · similarity 0.89)
- Sketch-Augmented Features Improve Learning Long-Range Dependencies in Graph Neural Networks (NeurIPS 2025 · similarity 0.88)
- Recurrent Distance Filtering for Graph Representation Learning (ICML 2024 · similarity 0.88)
- A Scalable and Effective Alternative to Graph Transformers (AAAI 2025 · similarity 0.88)
- What Improves the Generalization of Graph Transformers? A Theoretical Dive into the Self-attention and Positional Encoding (ICML 2024 · similarity 0.88)
- On Measuring Long-Range Interactions in Graph Neural Networks (ICML 2025 · 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.