Research map / Graph learning / Graph algorithms and clustering
A Gromov--Wasserstein Geometric View of Spectrum-Preserving Graph Coarsening
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 KGC(A) instead of VEGC-initialized coarsening | Graph regression on AQSOL · AQSOL | TestMAE: 1.390±0.005 -> 1.383±0.005 | Helped |
| Use KGC(A) instead of VEGC-initialized coarsening | Graph regression on ZINC · ZINC | TestMAE: 0.646±0.001 -> 0.641±0.003 | Helped |
| Initialize KGC with a baseline coarsening result | KGC graph classification · MUTAG, PROTEINS, Tumblr | accuracy: 81.90±2.74 -> 86.23±2.69; 63.45±0.83 -> 66.43±0.92; 52.52±2.81 -> 52.57±2.22 | Mixed |
| Initialize KGC with a baseline coarsening result | KGC graph classification · PTC, MSRC, IMDB | accuracy: 61.58±2.49 -> 57.25±2.16; 19.84±2.23 -> 17.17±2.91; 67.80±1.65 -> 69.20±1.37 | Mixed |
| Use weighted kernel K-means with averaging coarsening rather than projection coarsening | PTC GW-distance approximation · PTC | Frobenius error: 12.41±0.04 -> 4.45±0.03 | Helped |
| Use averaging instead of projection as the coarsening matrix | Graph distance matrix approximation · PTC | Frobenius error: 13.69±0. -> 5.31±0. | Helped |
| Initialize KGC with MGC | PTC GW-distance approximation · PTC | average squared GW distance: 0.03332 -> 0.03055 (c=0.3); 0.01255 -> 0.01609 (c=0.7) | Mixed |
| Choose normalized signless Laplacian similarity and minimize the kernel clustering objective | Graph coarsening · Tumblr | top-5 eigenvalue relative error | Helped |
Closest papers by meaning
- Taxonomy of reduction matrices for Graph Coarsening (NeurIPS 2025 · similarity 0.89)
- Spectral Graph Sparsification Preserves Representation Geometry in Graph Neural Networks (NeurIPS 2026 · similarity 0.89)
- Graph Coarsening with Message-Passing Guarantees (NeurIPS 2024 · similarity 0.88)
- Mitigating Over-Squashing in Graph Neural Networks by Spectrum-Preserving Sparsification (ICML 2025 · similarity 0.87)
- Structure-Aware Spectral Sparsification via Uniform Edge Sampling (NeurIPS 2025 · similarity 0.87)
- UGC: Universal Graph Coarsening (NeurIPS 2024 · 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.