Research map / Graph learning / GNN architectures
Learning Posterior Predictive Distributions for Node Classification from Synthetic Graph Priors
Jeongwhan Choi, Jongwoo Kim, Woosung Kang, Noseong Park
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 the cSBM component to ER-only prior (ablation) | NodePFN · Cora | accuracy: 80.62 -> 82.06 | Improved |
| add ER graphs to cSBM-only (full range) prior (ablation) | NodePFN · Cora | accuracy: 81.26 -> 82.06 | Improved |
| train on the full homophily range instead of 0.1-0.3 only (ablation) | NodePFN · Cora | accuracy: 79.89 -> 81.26 | Improved |
| mix ER and cSBM instead of only BA networks (ablation) | NodePFN · Cora | accuracy: 74.18 -> 82.06 | Improved |
| mix ER and cSBM instead of only BA networks (ablation) | NodePFN · Tolokers | accuracy: 74.63 -> 78.61 | Improved |
| mix ER and cSBM instead of only ER (ablation) | NodePFN · Wisconsin | accuracy: 80.39 -> 81.18 | Improved |
| add graph-aware dual-branch design over TabPFN (average, homophily) | NodePFN · 13 homophily graphs | accuracy: 55.62 -> 77.39 | Improved |
| add graph-aware dual-branch design over TabPFN (average, heterophily) | NodePFN · 10 heterophily graphs | accuracy: 53.47 -> 65.14 | Improved |
| use NodePFN instead of GAT on heterophily graphs | NodePFN · 10 heterophily graphs | accuracy: 59.11 -> 65.14 | Improved |
| use NodePFN instead of GCN | NodePFN · 23 benchmarks | accuracy: 66.63 -> 71.27 | Improved |
| scale synthetic prior graphs up to about 250,000 (ablation) | NodePFN · Texas | accuracy: 53% -> 76% | Improved |
| use NodePFN instead of GraphAny trained on Cora on Wisconsin-type data | NodePFN · Wisconsin | accuracy: 61.18 -> 79.22 | Improved |
Closest papers by meaning
- Beyond Fixed Depth: Adaptive Graph Neural Networks for Node Classification Under Varying Homophily (AAAI 2026 · similarity 0.90)
- GraphPFN: A Prior-Data Fitted Graph Foundation Model (ICML 2026 · similarity 0.90)
- Generative Graph Pattern Machine (NeurIPS 2025 · similarity 0.89)
- Label-free Node Classification on Graphs with Large Language Models (LLMs) (ICLR 2024 · similarity 0.89)
- Graph Classification via Reference Distribution Learning: Theory and Practice (NeurIPS 2024 · similarity 0.89)
- Diffusing to the Top: Boost Graph Neural Networks with Minimal Hyperparameter Tuning (ICLR 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.
