Research map / Graph learning / GNN architectures
Geodesic Graph Neural Network for Efficient 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 vertical geodesic representation to GCN | GCN · OGBL-PPA | Hits@100 %: 18.67 -> 45.92 | Helped |
| add shortest-path distance only vs GCN (ablation) | GCN · Cora link prediction | AUC: 81.79 -> 92.93 | Helped |
| add distance only vs GCN (ablation) | GCN · OGBL-PPA | Hits@100 %: 18.67 -> 20.39 | No clear effect |
| replace distance with full vertical geodesic (ablation) | GCN+DIST · OGBL-PPA | Hits@100 %: 20.39 -> 43.86 | Helped |
| add geodesic node degree to vertical geodesic (ablation) | GDGNN-Vert · OGBL-PPA | Hits@100 %: 43.86 -> 45.92 | Helped |
| add geodesic node degree to vertical geodesic (ablation) | GDGNN-Vert · Cora link prediction | AUC: 94.14 -> 94.56 | No clear effect |
| use full horizontal geodesic instead of partial (ablation) | 4-GDGNN-Hor · Cora link prediction | AUC: 92.73 -> 94.40 | Helped |
| add geodesic node degree to vertical geodesic (ablation) | GDGNN-Vert · OGBG-MOLHIV | AUC %: 77.9 -> 79.1 | Helped |
| increase cutoff distance from 3 to 4 (ablation) | GDGNN-Vert · OGBL-PPA | Hits@100 %: 45.92 -> 44.82 | Hurt |
| increase cutoff distance from 1 to 4 (ablation) | GDGNN-Vert · Cora link prediction | AUC: 82.46 -> 94.47 | Helped |
| use horizontal instead of vertical geodesic | GDGNN · OGBL-PPA | Hits@100 %: 45.92 -> 28.79 | Hurt |
| use GDGNN instead of subgraph method SEAL for efficiency | GDGNN-Vert · OGBL-PPA | running time: 2 hr 20 min -> 33 min | Helped |
Closest papers by meaning
- How Powerful are Graph Neural Networks? (ICLR 2019 · similarity 0.89)
- Distance Encoding: Design Provably More Powerful Neural Networks for Graph Representation Learning (NeurIPS 2020 · similarity 0.89)
- Graph Distillation with Eigenbasis Matching (ICML 2024 · similarity 0.89)
- Translating Subgraphs to Nodes Makes Simple GNNs Strong and Efficient for Subgraph Representation Learning (ICML 2024 · similarity 0.88)
- A Flexible, Equivariant Framework for Subgraph GNNs via Graph Products and Graph Coarsening (NeurIPS 2024 · similarity 0.88)
- Union Subgraph Neural Networks (AAAI 2024 · 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.