Research map / Learning theory and algorithms / Sample complexity
Estimating the Size of a Large Network and its Communities from a Random Sample
Lin Chen, Amin Karbasi, Forrest W. Crawford
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 PULSE posterior mean instead of NSUM | network size estimator · Erdos-Renyi p sweep | relative error: NSUM positive bias; PULSE within -1% to 1% | Improved |
| increase edge probability p from 0.1 to 0.9 (ablation) | PULSE · Erdos-Renyi | relative error variance | Improved |
| increase edge probability p (ablation) | NSUM · Erdos-Renyi | bias | No clear effect |
| increase network size N from 400 to 1000 at fixed sample size (ablation) | PULSE · Erdos-Renyi | relative error range: -0.7% to 0.7% -> -1.5% to 1.5% | Hurt |
| increase network size N (ablation) | NSUM · Erdos-Renyi | overestimation | Hurt |
| increase sample size |W| from 100 to 500 (ablation) | PULSE and NSUM · Erdos-Renyi N=1000 | IQR of relative error | Improved |
| increase sample size |W| from 100 to 500 (ablation) | NSUM · Erdos-Renyi N=1000 | bias | Improved |
| balanced type partition N1/N near 0.5 with half sampled (ablation) | PULSE · SBM K=2 N=200 | relative error | Improved |
| vary cohesion parameter epsilon from -0.3 to 0.3 (ablation) | PULSE vs NSUM · SBM K=2 N=850 | relative error | Mixed |
| increase number of types K (ablation) | PULSE and NSUM · SBM CRP N=200 | IQR of relative error | Hurt |
| increase sampling fraction n/N from 33% to 66% (ablation) | PULSE and NSUM · SBM CRP N=200 | relative error | Improved |
| PULSE vs NSUM at same K and sampling fraction | PULSE · SBM CRP N=200 | relative error | Improved |
Closest papers by meaning
- Determining the Number of Communities in Degree-corrected Stochastic Block Models (JMLR 2021 · similarity 0.88)
- Community models for networks observed through edge nominations (JMLR 2023 · similarity 0.87)
- Community Recovery in the Geometric Block Model (JMLR 2023 · similarity 0.86)
- Estimating the Number of Induced Subgraphs from Incomplete Data and Neighborhood Queries (AAAI 2021 · similarity 0.86)
- Network two-sample test for block models (NeurIPS 2025 · similarity 0.86)
- Optimal community detection in dense bipartite graphs (NeurIPS 2025 · similarity 0.84)
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.
