Research map / Graph learning / GNN architectures
Affinity-Aware Graph Networks
Ameya Velingker, Ali Kemal Sinop, Ira Ktena, Petar Veličković, Sreenivas Gollapudi
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 |
|---|---|---|---|
| add effective-resistance edge features to MPNN | MPNN · PNA test avg | log(MSE): -2.665 -> -2.779 | Improved |
| use hitting times instead of effective resistance as edge features | ER GNN · PNA test avg | log(MSE): -2.779 -> -2.816 | Improved |
| combine all ER features (ER, HT, node and edge embeddings) | ER GNN · PNA test avg | log(MSE): -2.779 -> -3.106 | Improved |
| add random node features to MPNN | MPNN · PNA test avg | log(MSE): -2.665 -> -2.490 | Hurt |
| apply random rotations to ER node embeddings (ablation) | MPNN + ER node embeddings · ogbg-molhiv | ROC-AUC: 77.77 -> 76.28 | Hurt |
| apply random rotations to ER node embeddings + HT (ablation) | MPNN + ER node embeddings + HT · ogbg-molhiv | ROC-AUC: 78.16 -> 79.13 | Improved |
| add hitting times to ER node embeddings (ablation) | MPNN + ER node embeddings · ogbg-molhiv | ROC-AUC: 77.77 -> 78.16 | Improved |
| add random features to MPNN (ablation) | MPNN · ogbg-molpcba (4 layers) | mAP: 27.75 -> 26.69 | Hurt |
| add ER features to MPNN + Noisy Nodes at 4 layers | MPNN + Noisy Nodes · ogbg-molpcba (4 layers) | mAP: 27.92 -> 28.11 | Improved |
| add ER features to MPNN + Noisy Nodes at 16 layers | MPNN + Noisy Nodes · ogbg-molpcba (16 layers) | mAP: 28.29 -> 28.28 | No clear effect |
| add ER scalar features to GAT | GAT · ogbn-arXiv | accuracy: 72.02 -> 72.14 | Improved |
| add hitting time features to GAT | GAT · ogbn-arXiv | accuracy: 72.14 -> 72.25 | Improved |
Closest papers by meaning
- Revisting Node Affinity Prediction In Temporal Graphs (ICLR 2026 · similarity 0.89)
- Learning Long Range Dependencies on Graphs via Random Walks (ICLR 2025 · similarity 0.89)
- Bridging Theory and Practice in Link Representation with Graph Neural Networks (NeurIPS 2025 · similarity 0.88)
- Towards Dynamic Message Passing on Graphs (NeurIPS 2024 · similarity 0.88)
- A Topological Perspective on Demystifying GNN-Based Link Prediction Performance (ICLR 2024 · similarity 0.88)
- Beyond Message Passing: Neural Graph Pattern Machine (ICML 2025 · similarity 0.87)
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.
