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 Bayesian Approach to Model Selection in Hierarchical Mixtures-of-Experts Architectures.

Martin A. Tanner1, Fengchun Peng, Robert A. Jacobs

  • 1Northwestern University, USA

Neural Networks : the Official Journal of the International Neural Network Society
|March 1, 1997
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

Commentaries on Viewpoint: Principles, insights, and potential pitfalls of the noninvasive determination of muscle oxidative capacity by near-infrared spectroscopy.

Journal of applied physiology (Bethesda, Md. : 1985)·2018
Same author

What determines visual cue reliability?

Trends in cognitive sciences·2002
Same journal

Exploiting audio-visual modalities in videos: Object detection via multi-stage bilateral coupling network.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

Reliability-aware modality completion with cross-modal distillation for federated learning with missing modalities.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

IGFD-Net: Illumination-guided frequency decoupling for polarization image fusion.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

Multiple-Strategies dung beetle optimizer and its applications in engineering optimization and bankruptcy prediction.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

Aggregating global-scale pixel-wise forgery cues within a graph.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

Finite-Time intermittent control for secure synchronization of Neutral-Type stochastic delayed neural networks under aperiodic DoS attacks.

Neural networks : the official journal of the International Neural Network Society·2026
See all related articles

Selecting the best statistical model is crucial for data analysis. This study introduces a novel approach for hierarchical mixtures-of-experts, favoring simpler, effective models through component pruning and Bayesian hypothesis testing.

Area of Science:

  • Machine Learning
  • Statistical Modeling
  • Artificial Intelligence

Background:

  • No single statistical model performs optimally across all tasks, necessitating a model selection process.
  • Hierarchical mixtures-of-experts (HME) architectures combine generalized linear models and finite mixture models for recursive data analysis.
  • Effective model selection is vital for accurate data summarization and task performance.

Purpose of the Study:

  • To present a new approach for addressing the model selection problem within HME architectures.
  • To develop methods for estimating component importance and pruning unnecessary parts of the HME structure.
  • To utilize Bayesian hypothesis testing for distinguishing informative inputs from nuisance inputs.

Main Methods:

  • Employs hierarchical mixtures-of-experts (HME) architectures utilizing a recursive "divide-and-conquer" strategy.

Related Experiment Videos

  • Estimates model parameter distributions using Markov chain Monte Carlo (MCMC) methodology.
  • Incorporates a component worth estimation for pruning and Bayesian hypothesis testing for input differentiation.

    • Main Results:

    • The proposed approach favors simpler HME architectures that adequately summarize data, aligning with Occam's razor.
    • Simulation results demonstrate the effectiveness of pruning unused components and identifying informative inputs.
    • The method successfully differentiates between useful and nuisance inputs, leading to more parsimonious models.

    Conclusions:

    • The presented approach offers an effective strategy for model selection in HME architectures.
    • It promotes the development of simpler, more interpretable, and computationally efficient models.
    • This methodology enhances the practical application of complex statistical models in diverse data analysis tasks.