Research map / Graph learning / GNN architectures
Weisfeiler and Leman go sparse: Towards scalable higher-order graph embeddings
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 local sparse tuple neighborhoods instead of global neighborhoods | δ-k-LWL graph kernels · Graph classification benchmarks | accuracy: almost 20% | Helped |
| Add LWL+ global-count information to local aggregation (ablation) | δ-2-LWL · IMDB-Multi | accuracy: 50.2 ± 0.6 -> 62.5 ± 1.0 | Helped |
| Add LWL+ global-count information to local aggregation (ablation) | δ-2-LWL · Enzymes | accuracy: 56.6 ± 1.2 -> 52.9 ± 1.4 | Hurt |
| Increase tuple order from 2 to 3 (ablation) | δ-k-LWL · Graph classification benchmarks excluding Enzymes | accuracy | Hurt |
| Use local rather than global tuple neighborhoods (ablation) | δ-2-LGNN · Zinc | MAE: 0.042 ± 0.003 -> 0.045 ± 0.006 | Hurt |
| Use local rather than global tuple neighborhoods (ablation) | δ-2-LGNN · Alchemy | MAE: 0.080 ± 0.001 -> 0.083 ± 0.001 | Hurt |
| Use local aggregation rather than global aggregation | δ-2-WL kernel · Enzymes test split | accuracy: 37.5 -> 56.6 | Helped |
| Use local sparse tuple neighborhoods | δ-k graph kernels · Enzymes | speedup: over ten times faster | Helped |
| Use local sparse tuple neighborhoods | δ-2-LGNN · Zinc | speedup: more than two times lower | Helped |
Closest papers by meaning
- Weisfeiler-Leman at the margin: When more expressivity matters (ICML 2024 · similarity 0.91)
- Generalizing Weisfeiler-Lehman Kernels to Subgraphs (ICLR 2025 · similarity 0.89)
- Aligning Transformers with Weisfeiler-Leman (ICML 2024 · similarity 0.88)
- Weisfeiler-Leman Is Incomplete on Simple Spectrum Graphs, so Canonicalize Them (NeurIPS 2026 · similarity 0.88)
- Weisfeiler and Leman Go Loopy: A New Hierarchy for Graph Representational Learning (NeurIPS 2024 · similarity 0.87)
- On the Power of the Weisfeiler-Leman Test for Graph Motif Parameters (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.