Research map / Graph learning / GNN architectures
Graph Neural Tangent Kernel: Fusing Graph Neural Networks with Graph Kernels
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 GNTK instead of the strongest baseline | Graph classification · COLLAB | accuracy: 81.0 $\pm$ 0.3 -> 83.6 $\pm$ 1.0 | Helped |
| Use GNTK instead of the strongest baseline | Graph classification · IMDB-B | accuracy: 75.1 $\pm$ 5.1 -> 76.9 $\pm$ 3.6 | Helped |
| Use GNTK instead of the strongest baseline | Graph classification · IMDB-M | accuracy: 52.3 $\pm$ 2.8 -> 52.8 $\pm$ 4.6 | Helped |
| Use GNTK instead of the strongest baseline | Graph classification · PTC | accuracy: 64.6 $\pm$ 7.0 -> 67.9 $\pm$ 6.9 | Helped |
| Use GNTK instead of the strongest baseline | Graph classification · NCI1 | accuracy: 86.0 $\pm$ 1.8 -> 84.2 $\pm$ 1.5 | Hurt |
| Use GNTK instead of the strongest baseline | Graph classification · MUTAG | accuracy: 92.6 $\pm$ 4.2 -> 90.0 $\pm$ 8.5 | Hurt |
| Use GNTK instead of the strongest baseline | Graph classification · PROTEINS | accuracy: 76.2 $\pm$ 2.8 -> 75.6 $\pm$ 4.2 | Hurt |
| Add jumping knowledge (ablation) | GNTK · NCI1 and IMDB-BINARY | accuracy | Helped |
| Increase the number of MLP layers (ablation) | GNTK · NCI1 and IMDB-BINARY | accuracy: ~ 0.8% | Helped |
| Increase the number of BLOCK operations (ablation) | GNTK · NCI1 | accuracy | Helped |
| Use sum aggregation (c_u=1) instead of average aggregation (ablation) | GNTK · Social network datasets | accuracy | Helped |
| Replace GIN with GNTK using the same architecture | IMDB-B · IMDB-B | runtime: 19 minutes -> 2 minutes | Helped |
Closest papers by meaning
- Temporal Graph Neural Tangent Kernel with Graphon-Guaranteed (NeurIPS 2024 · similarity 0.93)
- How Graph Neural Networks Learn: Lessons from Training Dynamics (ICML 2024 · similarity 0.92)
- Neural Tangent Kernels Motivate Cross-Covariance Graphs in Neural Networks (ICML 2024 · similarity 0.91)
- Graph Neural Networks Are More Than Filters: Revisiting and Benchmarking from A Spectral Perspective (ICLR 2025 · similarity 0.89)
- Exploring Consistency in Graph Representations: from Graph Kernels to Graph Neural Networks (NeurIPS 2024 · similarity 0.89)
- What do Graph Neural Networks learn? Insights from Tropical Geometry (NeurIPS 2024 · similarity 0.89)
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.