Research map / Graph learning / GNN architectures
NAFS: A Simple yet Tough-to-beat Baseline for 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 |
|---|---|---|---|
| Use NAFS-concat instead of AGE | Node clustering · Cora | ACC: +2.6% | Helped |
| Use NAFS-concat instead of AGE | Node clustering · Citeseer | NMI: +1.8% | Helped |
| Use NAFS-mean instead of AGE | Node clustering · PubMed | NMI: +3.8% | Helped |
| Use NAFS-mean instead of AGE | Node clustering · Wiki | ACC: +3.5% | Helped |
| Use NAFS-concat instead of ARGA | Link prediction · PubMed | AUC: 0.8% | Helped |
| Replace node-adaptive weights with single-hop selection (ablation) | NAFS · PubMed | NMI | Hurt |
| Replace node-adaptive weights with equal-weight averaging (ablation) | NAFS · PubMed | NMI | Hurt |
| Increase maximal smoothing steps while using equal-weight averaging (ablation) | NAFS · PubMed | NMI | Hurt |
| Add feature ensemble across smoothing operators instead of using r=0.3 only (ablation) | NAFS · PubMed | NMI | Helped |
| Choose mean, max, or concatenation as the ensemble strategy (ablation) | NAFS · Multiple datasets and tasks | NMI | Mixed |
| Increase edge sparsity and use NAFS-mean instead of GAE | PubMed link prediction · PubMed edge-removal setting | AUC: 1.0% to 3.7% | Helped |
| Increase feature sparsity and use NAFS-mean instead of GAE | PubMed link prediction · PubMed feature-removal setting | AP: 0.7% to 5.9% | Helped |
Closest papers by meaning
- Adaptive Node Feature Selection for Graph Neural Networks (ICML 2026 · similarity 0.91)
- Node Dependent Local Smoothing for Scalable Graph Learning (NeurIPS 2021 · similarity 0.90)
- A Fair Comparison of Graph Neural Networks for Graph Classification (ICLR 2020 · similarity 0.90)
- Subgraph Invariant Learning Towards Large-Scale Graph Node Classification (AAAI 2025 · similarity 0.88)
- Bootstrapping Heterogeneous Graph Representation Learning via Large Language Models: A Generalized Approach (AAAI 2025 · similarity 0.88)
- Node Identifiers: Compact, Discrete Representations for Efficient Graph Learning (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.