Research map / Graph learning / GNN architectures
A Variational Edge Partition Model for Supervised Graph Representation Learning
Yilin He, Chaojie Wang, Hao Zhang, Bo Chen, Mingyuan Zhou
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 |
|---|---|---|---|
| Learn task-relevant overlapping communities and partition edges for community-specific GNN aggregation | VEPM · Cora | accuracy: 83.7 -> 84.3 | Improved |
| Learn task-relevant overlapping communities and partition edges for community-specific GNN aggregation | VEPM · Pubmed | accuracy: 80.5 -> 82.4 | Improved |
| Learn task-relevant overlapping communities and partition edges for community-specific GNN aggregation | VEPM · MUTAG | accuracy: 92.6 ± 4.2 -> 93.6 ± 3.4 | Improved |
| Learn task-relevant overlapping communities and partition edges for community-specific GNN aggregation | VEPM · NCI1 | accuracy: 86.0 ± 1.8 -> 83.9 ± 1.8 | Hurt |
| Replace the learned edge partition with an even partition (ablation) | VEPM · IMDB-B | accuracy: 76.7 ± 3.1 -> 74.2 ± 4.3 | Hurt |
| Replace the learned edge partition with a fixed random partition (ablation) | VEPM · IMDB-B | accuracy: 76.7 ± 3.1 -> 60.5 ± 4.7 | Hurt |
| Replace the GNN representation composer with an FC layer (ablation) | VEPM with meaningful partitions · PTC | accuracy: 75.6 ± 5.9 -> 68.7 ± 4.3 | Hurt |
| Increase edge-partition temperature on random partitions (ablation) | VEPM · MUTAG | accuracy: 79.2 ± 7.9 -> 89.8 ± 4.6 | Improved |
| Use subgraph-sampling-based scalable training (ablation) | VEPM · Cora | accuracy: 84.3 ± 0.1 -> 83.4 ± 0.1 | Hurt |
| Use subgraph-sampling-based scalable training (ablation) | VEPM · Cora | epoch duration: 0.19s -> 0.02s | Improved |
| Add community affiliation scores to hand-crafted input features (ablation) | VEPM · IMDB-B | accuracy: 80.3 ± 2.0 -> 76.7 ± 3.1 | No clear effect |
| Add unsupervised pretraining before supervised finetuning (ablation) | VEPM · IMDB-B | accuracy: 61.0 ± 3.6 -> 76.7 ± 3.1 | Improved |
Closest papers by meaning
- Supervised Community Detection with Line Graph Neural Networks (ICLR 2019 · similarity 0.90)
- Integrating Co-Training with Edge Discrimination to Enhance Graph Neural Networks Under Heterophily (AAAI 2025 · similarity 0.90)
- Graph Neural Networks Powered by Encoder Embedding for Improved Node Learning (TPAMI 2026 · similarity 0.90)
- Improving Graph Neural Networks by Learning Continuous Edge Directions (ICLR 2025 · similarity 0.89)
- NLGT: Neighborhood-based and Label-enhanced Graph Transformer Framework for Node Classification (AAAI 2025 · similarity 0.89)
- Graph Classification via Reference Distribution Learning: Theory and Practice (NeurIPS 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.
