Research map / Graph learning / GNN architectures
Design Space for Graph Neural Networks
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 |
|---|---|---|---|
| Add batch normalization (ablation) | GNN designs · 32 GNN tasks | average rank: 1.44 -> 1.15 | Helped |
| Use dropout on node feature dimensions (ablation) | GNN designs · 32 GNN tasks | performance | No clear effect |
| Use PReLU activation (ablation) | GNN designs · 32 GNN tasks | performance | Helped |
| Use sum aggregation (ablation) | GNN designs · 32 GNN tasks | performance | Helped |
| Add skip connections instead of stacking layers (ablation) | GNN designs · 32 GNN tasks | performance | Helped |
| Use concatenation skip connections instead of residual sum connections (ablation) | GNN designs · 32 GNN tasks | performance | Helped |
| Choose the number of message-passing layers (ablation) | GNN designs · 32 GNN tasks | performance | Mixed |
| Choose the number of pre-processing or post-processing layers (ablation) | GNN designs · 32 GNN tasks | performance | Mixed |
| Use batch size 32 (ablation) | GNN designs · 32 GNN tasks | ranking | Helped |
| Use learning rate 0.01 (ablation) | GNN designs · 32 GNN tasks | ranking | Helped |
| Use Adam instead of SGD (ablation) | GNN designs · 32 GNN tasks | performance | Helped |
| Train for more epochs (ablation) | GNN designs · 32 GNN tasks | performance | Helped |
Closest papers by meaning
- Extending the Design Space of Graph Neural Networks by Rethinking Folklore Weisfeiler-Lehman (NeurIPS 2023 · similarity 0.90)
- Can Classic GNNs Be Strong Baselines for Graph-level Tasks? Simple Architectures Meet Excellence (ICML 2025 · similarity 0.90)
- Graph Neural Networks Are More Than Filters: Revisiting and Benchmarking from A Spectral Perspective (ICLR 2025 · similarity 0.88)
- On the Interplay between Graph Structure and Learning Algorithms in Graph Neural Networks (ICML 2025 · similarity 0.88)
- Classic GNNs are Strong Baselines: Reassessing GNNs for Node Classification (NeurIPS 2024 · similarity 0.88)
- How Graph Neural Networks Learn: Lessons from Training Dynamics (ICML 2024 · 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.