Research map / Probabilistic and scientific modeling / Variational and generative sampling
Improving Structure MCMC for Bayesian Networks through Markov Blanket Resampling
Chengwei Su, Mark E. Borsuk
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 MBR move to structure MCMC instead of REV move | structure MCMC · ALARM, HEPAR II, GENS2 (m=1000) | iterations to highest scoring structure: approximately half the number of iterations of REV and one tenth of the traditional sampler | Improved |
| add MBR move to structure MCMC | structure MCMC · ALARM (m=50 to 1000) | convergence reliability: MBR and REV converge approximately equally reliably | Improved |
| add MBR move instead of REV | structure MCMC · ALARM (m=100), 10 runs | AUROC 0.01 paired t-test: t = 7.78, p = 2.74e-05 | Improved |
| add MBR move instead of REV | structure MCMC · ALARM (m=100), 10 runs | AUROC 1 paired t-test: t = 3.52, p = 0.0065 | Improved |
| MBR/REV versus traditional move at m>=250 | structure MCMC · ALARM (m>=250) | AUROC | No clear effect |
| add MBR move on Hailfinder | structure MCMC · Hailfinder (m=500 to 1500) | AUROC | Mixed |
| increase MBR move frequency above 1/15 (ablation) | MBR structure MCMC · ALARM | sampling efficiency: pm=1/15 chosen | Hurt |
| add MBR move instead of REV | structure MCMC · ALARM (m=1000), 10,000 iterations | running time (sec): 401.5 -> 407.3 | No clear effect |
Closest papers by meaning
- Structure Discovery in Bayesian Networks by Sampling Partial Orders (JMLR 2016 · similarity 0.90)
- Efficient Bayesian network structure learning via local Markov boundary search (NeurIPS 2021 · similarity 0.90)
- Extendable and Iterative Structure Learning Strategy for Bayesian Networks (ICLR 2025 · similarity 0.89)
- A Gibbs Sampler for Learning DAGs (JMLR 2016 · similarity 0.89)
- Minimal I-MAP MCMC for Scalable Structure Discovery in Causal DAG Models (ICML 2018 · similarity 0.89)
- Structure Learning with Adaptive Random Neighborhood Informed MCMC (NeurIPS 2023 · similarity 0.89)
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.
