Research map / Graph learning / GNN architectures
Beyond Real-world Benchmark Datasets: An Empirical Study of Node Classification with GNNs
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 |
|---|---|---|---|
| make class sizes imbalanced (largest class share 0.4 to 0.7) instead of balanced | GNNs for node classification · synthetic Cora-like graphs (GenCAT) | f1-macro | Hurt |
| use simple SGC instead of complex GNNs under class imbalance | node classification · imbalanced synthetic graphs | f1-macro | Helped |
| use heterophily-aware GNNs (H2GCN, FSGNN, GPRGNN) instead of MLP under strong heterophily | node classification · synthetic graphs, beta=8 | f1-macro | No clear effect |
| use heterophily-aware GNNs instead of GCN/GAT in heterophilic graphs | node classification · synthetic heterophilic graphs | f1-macro | Helped |
| make attribute-class correlation close to random (large gamma) | GNNs · synthetic graphs | f1-macro | Hurt |
| increase graph size (3000 to 15000 nodes) | GNNs · synthetic graphs | f1-macro | Hurt |
| increase graph size | MLP · synthetic graphs | f1-macro | Helped |
| increase graph size | GNNs · synthetic graphs | training time per epoch | Hurt |
Closest papers by meaning
- Classic GNNs are Strong Baselines: Reassessing GNNs for Node Classification (NeurIPS 2024 · similarity 0.93)
- A critical look at the evaluation of GNNs under heterophily: Are we really making progress? (ICLR 2023 · similarity 0.93)
- Demystifying Structural Disparity in Graph Neural Networks: Can One Size Fit All? (NeurIPS 2023 · similarity 0.93)
- GC4NC: A Benchmark Framework for Graph Condensation on Node Classification with New Insights (NeurIPS 2025 · similarity 0.91)
- Can Classic GNNs Be Strong Baselines for Graph-level Tasks? Simple Architectures Meet Excellence (ICML 2025 · similarity 0.91)
- Graph Neural Networks Are More Than Filters: Revisiting and Benchmarking from A Spectral Perspective (ICLR 2025 · similarity 0.91)
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.