Research map / Graph learning / GNN architectures
Decouple Graph Neural Networks: Train Multiple Simple GNNs Simultaneously Instead of One
Hongyuan Zhang, Yanan Zhu, Xuelong Li
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 backward training (ablation) | SGNN-FT · Cora node clustering | clustering accuracy: 0.6278 -> 0.7463 | Improved |
| add backward training (ablation) | SGNN-FT · Reddit node clustering | clustering accuracy: 0.5943 -> 0.7042 | Improved |
| add backward training (ablation) | SGNN-FT · Reddit node classification | test accuracy (%): 94.54 -> 95.10 | Improved |
| add backward training (ablation) | SGNN-FT · OGB-Products node classification | test accuracy (%): 68.10 -> 74.44 | Improved |
| add backward training (ablation) | SGNN-FT · OGB-Arxiv node classification | test accuracy (%): 65.25 -> 71.57 | Improved |
| add backward training (ablation) | SGNN-FT · Cora node classification | test accuracy (%): 81.0 -> 81.4 | Improved |
| use SGNN decoupled training instead of GAE | GAE · Cora node clustering, L=7 | clustering accuracy: 0.52 -> 0.68 | Improved |
| increase depth from 2 to 7 | SGNN · Cora node clustering | clustering accuracy: 0.75 -> 0.68 | Hurt |
| increase depth from 2 to 7 | GAE · Cora node clustering | clustering accuracy: 0.60 -> 0.52 | No clear effect |
| use GAE architecture for S2GC (GAE-S2GC vs S2GC) | S2GC · Cora node clustering | clustering accuracy: 0.6960 -> 0.6976 | No clear effect |
| use S2GC as base module of SGNN | SGNN · Cora node classification | test accuracy (%): 81.4 -> 83.8 | Improved |
| decouple L-layer GNN into SGNN | GCN · Cora node classification | test accuracy (%): 81.4 -> 81.4 | No clear effect |
Closest papers by meaning
- Mitigating Oversmoothing Through Reverse Process of GNNs for Heterophilic Graphs (ICML 2024 · similarity 0.87)
- Forward Learning of Graph Neural Networks (ICLR 2024 · similarity 0.87)
- Dynamic Rescaling for Training GNNs (NeurIPS 2024 · similarity 0.87)
- On Vanishing Gradients, Over-Smoothing, and Over-Squashing in GNNs: Bridging Recurrent and Graph Learning (NeurIPS 2025 · similarity 0.87)
- How Graph Neural Networks Learn: Lessons from Training Dynamics (ICML 2024 · similarity 0.87)
- Locality-Aware Graph Rewiring in GNNs (ICLR 2024 · 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.
