Research map / Graph learning / GNN architectures
KerGNNs: Interpretable 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 KerGNN-2 instead of PK | KerGNN · DD | test accuracy (%): 78.0$\pm$3.8 -> 78.9$\pm$3.5 | Helped |
| Use KerGNN-2-DRW instead of GNTK | KerGNN · NCI1 | test accuracy (%): 83.5$\pm$1.2 -> 82.8$\pm$1.8 | Hurt |
| Use KerGNN-3 instead of Powerful GNN | KerGNN · PROTEINS | test accuracy (%): 75.9$\pm$3.3 -> 76.5$\pm$3.9 | Helped |
| Use KerGNN-1 instead of GCKN | KerGNN · ENZYMES | test accuracy (%): 59.3$\pm$5.6 -> 62.1$\pm$5.5 | Helped |
| Use KerGNN-1 instead of GNTK | KerGNN · IMDB-B | test accuracy (%): 75.9$\pm$3.1 -> 74.4$\pm$4.3 | Hurt |
| Use KerGNN-1 instead of GNTK | KerGNN · IMDB-M | test accuracy (%): 52.2$\pm$4.2 -> 51.6$\pm$3.1 | Hurt |
| Use KerGNN-2-DRW instead of 1-2-3 GNN | KerGNN · REDDIT-B | test accuracy (%): 91.1$\pm$2.1 -> 89.5$\pm$1.6 | Hurt |
| Use KerGNN-2-DRW instead of WL-sub | KerGNN · COLLAB | test accuracy (%): 77.5$\pm$2.4 -> 75.1$\pm$2.3 | Hurt |
| Use deep random walk kernel instead of standard random walk kernel (ablation) | KerGNN-2 · NCI1 | test accuracy (%): 76.3$\pm$2.6 -> 82.8$\pm$1.8 | Helped |
| Use deep random walk kernel instead of standard random walk kernel (ablation) | KerGNN-2 · DD | test accuracy (%): 78.9$\pm$3.5 -> 77.0$\pm$4.4 | Hurt |
| Use deep random walk kernel instead of standard random walk kernel (ablation) | KerGNN-2 · PROTEINS | test accuracy (%): 75.5$\pm$4.6 -> 76.1$\pm$4.1 | Mixed |
| Increase KerGNN depth from 1 to 3 layers (ablation) | KerGNN · NCI1 | test accuracy (%): 74.3$\pm$2.2 -> 80.5$\pm$1.9 | Helped |
Closest papers by meaning
- Convolutional Kernel Networks for Graph-Structured Data (ICML 2020 · similarity 0.91)
- Adaptive Kernel Graph Neural Network (AAAI 2022 · similarity 0.91)
- Exploring Consistency in Graph Representations: from Graph Kernels to Graph Neural Networks (NeurIPS 2024 · similarity 0.90)
- How Graph Neural Networks Learn: Lessons from Training Dynamics (ICML 2024 · similarity 0.89)
- Learning to Approximate Adaptive Kernel Convolution on Graphs (AAAI 2024 · similarity 0.89)
- Graph Classification via Reference Distribution Learning: Theory and Practice (NeurIPS 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.