Research map / Graph learning / Graph algorithms and clustering
Rethinking Efficient Graph Coarsening via a Non-Selfishness Principle
Xu Bai, Bin Lu, Kun Zhang, Shengbo Chen, Xinbing Wang, Chenghu Zhou, Meng Jin
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 NOPE* local-isotropy approximation instead of exact NOPE | NOPE · Citeseer coarsening r=0.5 | runtime: 0.97s -> 0.35s | Improved |
| use NOPE* instead of NOPE (ablation) | NOPE · Ogb-Arxiv coarsening r=0.5 | runtime: 5min12.79s -> 1min4.38s | Improved |
| use NOPE* instead of NOPE (ablation) | NOPE · Book coarsening r=0.5 | runtime: 44min48.80s -> 4min29.46s | Improved |
| use NOPE* instead of NOPE at high coarsening ratio (ablation) | NOPE · coarsening ratio r=0.7 case study | label prediction | Hurt |
| use NOPE* approximation (ablation) | NOPE · Citeseer LLM node classification r=0.5 | accuracy: 0.6206 -> 0.6018 | Hurt |
| use NOPE* approximation (ablation) | NOPE · Book LLM node classification r=0.5 | accuracy: 0.9170 -> 0.9244 | Improved |
| add interference-aware (non-selfish) merging instead of selfish pairwise merging | greedy graph coarsening · Dirichlet Energy during coarsening | Dirichlet Energy | Improved |
| use NOPE instead of heuristic sampling (Random/Degree/RAG) | LLM node classification · Products (subset) r=0.5 | accuracy: 0.5789 -> 0.6783 | Improved |
| use NOPE over UGC | LLM node classification · Ogb-Arxiv r=0.5 | accuracy: 0.3482 -> 0.3808 | Improved |
| use NOPE instead of A-CM | LLM node classification · Ogb-Arxiv r=0.5 | accuracy: 0.3925 -> 0.3808 | Hurt |
| raise coarsening rate for NOPE | NOPE · runtime at high r | runtime | Hurt |
| coarsen with NOPE and train GNN on coarsened graph | GCN/GIN/SAGE · node classification r=0.5 | accuracy | No clear effect |
Closest papers by meaning
- Featured Graph Coarsening with Similarity Guarantees (ICML 2023 · similarity 0.89)
- Fast and Scalable Hashing-Based Universal Graph Coarsening (TPAMI 2026 · similarity 0.89)
- UGC: Universal Graph Coarsening (NeurIPS 2024 · similarity 0.87)
- Graph Coarsening with Message-Passing Guarantees (NeurIPS 2024 · similarity 0.85)
- THESAURUS: Contrastive Graph Clustering by Swapping Fused Gromov-Wasserstein Couplings (AAAI 2025 · similarity 0.85)
- Equivalence is All: A Unified View for Self-supervised Graph Learning (ICML 2025 · similarity 0.85)
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.
