Research map / Graph learning / GNN architectures
Subgraph Neural Networks
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 |
|---|---|---|---|
| Use subgraph-level message passing with property-aware channels rather than baseline subgraph representations | Subgraph classification · synthetic datasets | Micro-F1: SubGNN 0.919\std{0.016} vs strongest baseline 0.803\std{0.039} on density; 0.629\std{0.039} vs 0.423\std{0.057} on cut r | Helped |
| Use SubGNN with GIN node embeddings | Subgraph classification · ppi-bp | Micro-F1: 0.398\std{0.058} -> 0.599\std{0.024} | Helped |
| Use SubGNN with GraphSAINT node embeddings | Subgraph classification · hpo-neuro | Micro-F1: 0.535\std{0.032} -> 0.644\std{0.019} | Helped |
| Use SubGNN with GIN node embeddings | Subgraph classification · hpo-metab | Micro-F1: 0.452\std{0.025} -> 0.537\std{0.023} | Helped |
| Use SubGNN with GraphSAINT node embeddings instead of the strongest baseline | Subgraph classification · em-user | Micro-F1: 0.859\std{0.014} -> 0.816\std{0.040} | Hurt |
| Use only the structure channel instead of the full channel combination (ablation) | SubGNN · density | Micro-F1: 0.894\std{0.025} -> 0.919\std{0.016} | Helped |
| Use only the structure channel instead of the full channel combination (ablation) | SubGNN · cut ratio | Micro-F1: 0.458\std{0.101} -> 0.629\std{0.039} | Helped |
| Use only the structure channel instead of the full channel combination (ablation) | SubGNN · coreness | Micro-F1: 0.659\std{0.092} -> 0.663\std{0.058} | No clear effect |
| Use only the position channel instead of the full channel combination (ablation) | SubGNN · component | Micro-F1: 0.726\std{0.120} -> 0.958\std{0.098} | Helped |
| Replace sum readout with attention over component representations (ablation) | SubGNN · subgraph classification | performance | No clear effect |
Closest papers by meaning
- Translating Subgraphs to Nodes Makes Simple GNNs Strong and Efficient for Subgraph Representation Learning (ICML 2024 · similarity 0.92)
- A Flexible, Equivariant Framework for Subgraph GNNs via Graph Products and Graph Coarsening (NeurIPS 2024 · similarity 0.92)
- Union Subgraph Neural Networks (AAAI 2024 · similarity 0.91)
- Implicit Subgraph Neural Network (ICML 2025 · similarity 0.90)
- An Efficient Subgraph-Inferring Framework for Large-Scale Heterogeneous Graphs (AAAI 2024 · similarity 0.90)
- Subgraph Invariant Learning Towards Large-Scale Graph Node Classification (AAAI 2025 · similarity 0.90)
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.