Research map / Graph learning / Graph algorithms and clustering
Wasserstein Weisfeiler-Lehman 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 |
|---|---|---|---|
| replace summed RBF kernel over node embeddings with Wasserstein distance (ablation) | WWL · ENZYMES (continuous) | accuracy (%): 68.43 -> 73.25 | Helped |
| replace summed RBF kernel with Wasserstein distance (ablation) | WWL · BZR (continuous) | accuracy (%): 80.96 -> 84.42 | Helped |
| replace summed RBF kernel with Wasserstein distance (ablation) | WWL · BZR-MD (continuous) | accuracy (%): 69.13 -> 69.76 | No clear effect |
| replace WL aggregation by sum with Wasserstein transport on WL embeddings | WWL · PTC-MR | accuracy (%): 61.21 -> 66.31 | Helped |
| replace optimal assignment (WL-OA) with optimal transport (WWL) | WWL · NCI1 | accuracy (%): 86.08 -> 85.75 | No clear effect |
| replace optimal assignment (WL-OA) with optimal transport (WWL) | WWL · PROTEINS (categorical) | accuracy (%): 76.37 -> 74.28 | Hurt |
| replace WL-OA with WWL | WWL · ENZYMES (categorical) | accuracy (%): 58.97 -> 59.13 | No clear effect |
| replace hash graph kernel HGK-WL with WWL | WWL · ENZYMES (continuous) | accuracy (%): 63.04 -> 73.25 | Helped |
| replace HGK-WL with WWL | WWL · COX2 (continuous) | accuracy (%): 78.13 -> 78.29 | No clear effect |
| replace Krein SVM with standard SVM for continuous WWL (ablation) | WWL · continuous attribute datasets | accuracy (%) | No clear effect |
Closest papers by meaning
- Isolation Graph Kernel (AAAI 2021 · similarity 0.93)
- Wasserstein Embedding for Graph Learning (ICLR 2021 · similarity 0.90)
- Graph Filtration Kernels (AAAI 2022 · similarity 0.90)
- Rethinking Kernel Methods for Node Representation Learning on Graphs (NeurIPS 2019 · similarity 0.90)
- DHAKR: Learning Deep Hierarchical Attention-Based Kernelized Representations for Graph Classification (AAAI 2025 · similarity 0.87)
- Statistical and Computational Guarantees of Kernel Max-Sliced Wasserstein Distances (ICML 2025 · similarity 0.86)
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.