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

Randomness in generalization ability: a source to improve it.

D Sarkar1

  • 1Dept. of Math. and Comput. Sci., Miami Univ., Coral Gables, FL.

IEEE Transactions on Neural Networks
|January 1, 1996
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

Estimating Reaction Rate Constants from Impedance Spectra: Combining Microkinetic Modeling and Experiments of the Oxygen Evolution Reaction.

The journal of physical chemistry. C, Nanomaterials and interfaces·2026
Same author

Long-Range Transverse-Momentum Correlations and Radial Flow in Pb-Pb Collisions at the LHC.

Physical review letters·2026
Same author

Advancing neural interfaces: a framework for the fabrication and characterization of freestanding micro-nanodevices.

Microsystems & nanoengineering·2026
Same author

Search for Quasiparticle Scattering in the Quark-Gluon Plasma with Jet Splittings in pp and Pb-Pb Collisions at sqrt[s_{NN}]=5.02  TeV.

Physical review letters·2025
Same author

First Measurement of A=4 Hypernuclei and Antihypernuclei at the LHC.

Physical review letters·2025
Same author

Experimental realization of a helical magnetic structure at Ni/Gd interfaces at room temperature.

Physical chemistry chemical physics : PCCP·2025
Same journal

Universal perceptron and DNA-like learning algorithm for binary neural networks: LSBF and PBF implementations.

IEEE transactions on neural networks·2013
Same journal

Guest editorial: special section on white box nonlinear prediction models.

IEEE transactions on neural networks·2011
Same journal

Data-based fault-tolerant control of high-speed trains with traction/braking notch nonlinearities and actuator failures.

IEEE transactions on neural networks·2011
Same journal

Guest editorial: special section on data-based control, modeling, and optimization.

IEEE transactions on neural networks·2011
Same journal

Neural network-based multiple robot simultaneous localization and mapping.

IEEE transactions on neural networks·2011
Same journal

Data-driven model-free adaptive control for a class of MIMO nonlinear discrete-time systems.

IEEE transactions on neural networks·2011
See all related articles

This study investigates randomness in feedforward artificial neural networks (FFANNs) generalization ability. A novel measurement method and a voting model demonstrate that combining multiple FFANNs can improve generalization performance.

Area of Science:

  • Artificial Intelligence
  • Machine Learning
  • Computational Neuroscience

Background:

  • Feedforward artificial neural networks (FFANNs) are widely used due to their simplicity and effectiveness.
  • Generalization ability is a critical performance metric for FFANNs, impacting their long-term utility.
  • Existing challenges in FFANN training, such as long learning times and local minima, are less concerning than generalization issues.

Purpose of the Study:

  • To investigate and quantify the randomness in the generalization ability of FFANNs.
  • To introduce a novel method for measuring the generalization ability of learning systems.
  • To develop a model for predicting the generalization ability of ensembles of FFANNs.

Main Methods:

  • A new metric was defined to measure the degree of randomness in FFANN generalization.

Related Experiment Videos

  • A 'voting model' was developed to predict the generalization ability of multiple combined FFANNs.
  • Analysis of VC-dimension was performed on the voting network model.

    • Main Results:

    • A novel method for measuring generalization ability was successfully defined and applied.
    • The study identified randomness in the generalization ability of FFANNs for specific problems.
    • The voting model demonstrated that increasing the number of networks improves generalization if individual network accuracy exceeds 50%.

    Conclusions:

    • Ensembling multiple FFANNs, using a voting model, can effectively mitigate randomness and enhance generalization ability.
    • The VC-dimension of the voting network model shows a potential monotonic increase with the number of constituent networks.
    • The findings provide a framework for improving the reliability and performance of FFANNs through ensemble methods.