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

Comparing Bayesian neural network algorithms for classifying segmented outdoor images.

F Vivarelli1, C K Williams

  • 1The Knowledge Lab-NCR Financial Solutions, London, UK. francesco.vivarelli@ncr.com

Neural Networks : the Official Journal of the International Neural Network Society
|June 20, 2001
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

Author Correction: Universal DNA methylation age across mammalian tissues.

Nature aging·2023
Same author

Universal DNA methylation age across mammalian tissues.

Nature aging·2023
Same author

Improving the solubility of pseudo-hydrophobic Alzheimer's Disease medicinal chemicals through co-crystal formulation.

bioRxiv : the preprint server for biology·2023
Same author

Efficacy of a new delivery system based on solid lipid microparticles for the oral administration of the non-conventional antioxidant IAC on a diabetes mouse model.

Journal of endocrinological investigation·2018
Same author

'Switch' catalysis: from monomer mixtures to sequence-controlled block copolymers.

Philosophical transactions. Series A, Mathematical, physical, and engineering sciences·2017
Same author

Catalysts for CO2/epoxide ring-opening copolymerization.

Philosophical transactions. Series A, Mathematical, physical, and engineering sciences·2016

This study compares two Bayesian neural network training methods for outdoor scene region labelling. The evidence framework and Markov Chain Monte Carlo methods showed comparable performance in empirical evaluations.

Area of Science:

  • Computer Vision
  • Machine Learning
  • Artificial Intelligence

Background:

  • Region labelling of outdoor scenes is crucial for environmental understanding.
  • Bayesian methods offer a principled approach to neural network training, quantifying uncertainty.
  • Existing methods may lack robust uncertainty quantification for complex scene data.

Purpose of the Study:

  • To investigate and compare two Bayesian training methods for neural networks in outdoor scene region labelling.
  • To evaluate the performance of the evidence framework and Markov Chain Monte Carlo (MCMC) methods.
  • To explore the utility of Automatic Relevance Determination (ARD) for feature selection in this context.

Main Methods:

  • Neural networks were trained using the evidence framework (MacKay, 1992) and a Markov Chain Monte Carlo (MCMC) method (Neal, 1996).

Related Experiment Videos

  • Performance was assessed by evaluating empirical learning curves.
  • The Automatic Relevance Determination (ARD) method was employed for input feature selection.

    • Main Results:

    • Both Bayesian training methods demonstrated comparable performance in region labelling tasks.
    • Empirical learning curves provided insights into the training dynamics of each method.
    • ARD facilitated effective input feature selection, potentially improving model efficiency.

    Conclusions:

    • Bayesian neural network training, using either the evidence framework or MCMC, is effective for outdoor scene region labelling.
    • The choice between these Bayesian methods may depend on specific application requirements and computational resources.
    • Feature selection using ARD can enhance the performance and interpretability of these models.