Research map / Learning theory and algorithms / Clustering
Learning Big Gaussian Bayesian Networks: Partition, Estimation and Fusion
Jiaying Gu, Qing Zhou
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 |
|---|---|---|---|
| add the fusion step after partition and estimation (ablation) | PEF-CCDr · ANDES(5), s0=1859, sb=169 | SHD: 1377.5 -> 626.3 | Improved |
| add the fusion step after partition and estimation (ablation) | PEF-CCDr · PATHFINDER(5), s0=975, sb=0 | SHD: 1145.8 -> 1003.6 | Improved |
| add the fusion step after partition and estimation (ablation) | PEF-PC · ANDES(5), s0=1690, sb=0 | SHD: 680.1 -> 315.3 | Improved |
| add the fusion step after partition and estimation (ablation) | PEF-PC · LINK(5), s0=6188, sb=563 | SHD: 3627.6 -> 1545.6 | Improved |
| add the fusion step after partition and estimation (ablation) | PEF-PC · ANDES(5), s0=1690, sb=0 | Jaccard index: 0.516 -> 0.796 | Improved |
| apply partition-estimation-fusion divide-and-conquer instead of CCDr on the whole data | CCDr · ANDES(5), s0=1690, sb=0 | SHD: 966.2 -> 653.3 | Improved |
| apply partition-estimation-fusion divide-and-conquer instead of PC on the whole data | PC · ANDES(5), s0=1690, sb=0 | SHD: 567.1 -> 315.3 | Improved |
| apply partition-estimation-fusion divide-and-conquer instead of CCDr on the whole data | CCDr · MUNIN(9, 0) | total running time (min): 542.56 -> 5.15 | Improved |
| apply partition-estimation-fusion divide-and-conquer instead of PC on the whole data | PC · PATHFINDER(5, 0) | total running time (min): 3.50 -> 1.89 | Improved |
| add between-subgraph edges (increase s_b) (ablation) | PEF-CCDr · ANDES(5) | Jaccard index: 0.576 -> 0.634 | No clear effect |
| use PC instead of CCDr in the E-step (ablation) | PEF framework · ANDES(5), s0=1690, sb=0 | SHD: 653.3 -> 315.3 | Improved |
| increase PC maximal conditioning set size m.max from 3 to 5 (ablation) | PC · Simulated networks | accuracy and time | Hurt |
Closest papers by meaning
- Extendable and Iterative Structure Learning Strategy for Bayesian Networks (ICLR 2025 · similarity 0.90)
- Learning Gaussian DAGs from Network Data (JMLR 2024 · similarity 0.90)
- Learning Identifiable Gaussian Bayesian Networks in Polynomial Time and Sample Complexity (NeurIPS 2017 · similarity 0.89)
- Parallel Bayesian Network Structure Learning (ICML 2018 · similarity 0.89)
- Structure Learning in Bayesian Networks of a Moderate Size by Efficient Sampling (JMLR 2016 · similarity 0.88)
- Distribution Learning Meets Graph Structure Sampling (NeurIPS 2025 · 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.
