Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

A Bootstrap Metropolis-Hastings Algorithm for Bayesian Analysis of Big Data.

Faming Liang1, Jinsu Kim2, Qifan Song3

  • 1Professor, Department of Biostatistics, University of Florida, Gainesville, FL 32611.

Technometrics : a Journal of Statistics for the Physical, Chemical, and Engineering Sciences
|October 17, 2017
PubMed
Summary
This summary is machine-generated.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Task-tailored Pre-processing: Fair Downstream Supervised Learning.

IEEE transactions on pattern analysis and machine intelligence·2026
Same author

Advances in root phenotyping: high-throughput imaging, computational tools and integrative approaches for crop improvement.

Journal of experimental botany·2026
Same author

Dietary management of pediatric patients with kidney disease: recommendations by the Korean Society of Pediatric Nephrology and the Korean Society of Clinical Nutrition.

Clinical nutrition research·2026
Same author

Fast Value Tracking for Deep Reinforcement Learning.

... International Conference on Learning Representations·2026
Same author

Causal-StoNet: Causal Inference for High-Dimensional Complex Data.

... International Conference on Learning Representations·2026
Same author

The role of exosomal miRNA-125b derived from colon cancer-associated fibroblasts in skeletal muscle cachexia.

PloS one·2026
Same journal

Bayesian Model Calibration and Sensitivity Analysis for Oscillating Biological Experiments.

Technometrics : a journal of statistics for the physical, chemical, and engineering sciences·2025
Same journal

Spatiotemporal Interactive Modeling of Event-based Dynamic Networks.

Technometrics : a journal of statistics for the physical, chemical, and engineering sciences·2025
Same journal

Note on the Equivalence of Orthogonalizing EM and Proximal Gradient Descent.

Technometrics : a journal of statistics for the physical, chemical, and engineering sciences·2025
Same journal

Anomaly Detection in Large-Scale Networks With Latent Space Models.

Technometrics : a journal of statistics for the physical, chemical, and engineering sciences·2024
Same journal

Robust Low-rank Tensor Decomposition with the <math></math> Criterion.

Technometrics : a journal of statistics for the physical, chemical, and engineering sciences·2024
Same journal

A Sharper Computational Tool for L<sub>2</sub>E Regression.

Technometrics : a journal of statistics for the physical, chemical, and engineering sciences·2023
See all related articles

We introduce the bootstrap Metropolis-Hastings (BMH) algorithm, a novel method enabling powerful Markov chain Monte Carlo (MCMC) techniques for big data analysis by utilizing parallel bootstrap samples. This approach makes complex data analysis feasible for large datasets.

Area of Science:

  • Computational Statistics
  • Data Science
  • Machine Learning

Background:

  • Markov chain Monte Carlo (MCMC) methods are powerful for complex data but computationally intensive.
  • Traditional MCMC methods are unsuitable for big data due to high computational demands and full dataset scans per iteration.

Purpose of the Study:

  • To propose a novel algorithm, bootstrap Metropolis-Hastings (BMH), for adapting MCMC methods to big data analysis.
  • To provide a general framework for making computationally intensive MCMC methods feasible for large-scale datasets.

Main Methods:

  • The bootstrap Metropolis-Hastings (BMH) algorithm replaces the full data log-likelihood with a Monte Carlo average from parallel bootstrap samples.
  • The BMH algorithm exhibits an embarrassingly parallel structure, avoiding repeated full dataset scans.
Keywords:
Big DataBootstrapMarkov Chain Monte CarloMetropolis-HastingsParallel Computing

Related Experiment Videos

  • BMH can be extended, demonstrated by the tempering BMH algorithm combining parallel tempering with BMH.

    • Main Results:

    • The BMH algorithm is shown to be feasible for big data problems.
    • BMH can be more efficient than divide-and-combine methods, integrating full data information in a single run.
    • The flexibility of BMH is demonstrated through its use in advanced MCMC algorithms, model selection, and optimization.

    Conclusions:

    • The bootstrap Metropolis-Hastings (BMH) algorithm offers a general and flexible framework for big data analysis using MCMC methods.
    • BMH enhances computational efficiency and scalability, making complex statistical analysis accessible for large datasets.
    • BMH serves as a foundational block for developing advanced, big data-compatible MCMC techniques.