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

Variational Bayes for continuous hidden Markov models and its application to active learning.

Shihao Ji1, Balaji Krishnapuram, Lawrence Carin

  • 1Department of Electrical and Computer Engineering, Duke University, Box 90291, Durham, NC 27708-0291, USA. shji@ee.duke.edu

IEEE Transactions on Pattern Analysis and Machine Intelligence
|March 29, 2006
PubMed
Summary

This study introduces a variational Bayes (VB) framework for continuous hidden Markov models (CHMMs) applied to active learning. The VB approach enhances model confidence, significantly reducing labeling needs compared to random selection.

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

Gradient Importance Learning for Incomplete Observations.

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

Fronto-Thalamo-Cerebellar Circuitry in Predicting Cognition and Behavior of ABCD Adolescents.

Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Annual International Conference·2025
Same author

Capturing Actionable Dynamics with Structured Latent Ordinary Differential Equations.

Proceedings of machine learning research·2025
Same author

Enabling Counterfactual Survival Analysis with Balanced Representations.

ACM CHIL 2021 : proceedings of the 2021 ACM Conference on Health, Inference, and Learning : April 8-9, 2021, Virtual Event. ACM Conference on Health, Inference, and Learning (2021 : Online)·2025
Same author

Detecting implicit biases of large language models with Bayesian hypothesis testing.

Scientific reports·2025
Same author

A machine learning algorithm improves the diagnostic accuracy of the histologic component of antibody mediated rejection (AMR-H) in cardiac transplant endomyocardial biopsies.

Cardiovascular pathology : the official journal of the Society for Cardiovascular Pathology·2024

Area of Science:

  • Machine Learning
  • Statistical Modeling
  • Artificial Intelligence

Background:

  • Traditional training procedures for continuous hidden Markov models (CHMMs) provide point estimates of parameters.
  • Variational Bayes (VB) offers a probabilistic approach, yielding a full posterior distribution of model parameters.
  • This probabilistic estimation is crucial for assessing confidence, especially with limited training data.

Purpose of the Study:

  • To introduce a variational Bayes (VB) framework for learning continuous hidden Markov models (CHMMs).
  • To integrate this VB framework into active learning strategies for efficient data labeling.
  • To evaluate the effectiveness of different active learning algorithms in reducing labeling requirements.

Main Methods:

  • Developed a variational Bayes (VB) framework for continuous hidden Markov models (CHMMs).

Related Experiment Videos

  • Implemented active learning algorithms: Query-by-Committee (QBC), maximum expected information gain, and error reduction.
  • Acquired labels for feature vectors that maximally reduce model parameter uncertainty.

    • Main Results:

    • All tested active learning methods significantly decreased the amount of required labeling compared to random selection.
    • The VB framework provides a measure of confidence in learned CHMM parameters, beneficial for small datasets.
    • Demonstrated the efficacy of VB-enhanced active learning on both synthetic and measured data.

    Conclusions:

    • The proposed variational Bayes framework effectively supports active learning for continuous hidden Markov models.
    • Active learning strategies, particularly those leveraging parameter uncertainty, substantially reduce data labeling costs.
    • This approach offers a robust method for building accurate CHMMs with minimal labeled data.