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

Direction-dependent learning approach for radial basis function networks.

Puneet Singla1, Kamesh Subbarao, John L Junkins

  • 1Department of Aerospace Engineering, Texas A&M University, College Station, TX 77843, USA. puneet@neo.tamu.edu

IEEE Transactions on Neural Networks
|February 7, 2007
PubMed
Summary

New methods for Gaussian basis functions improve input-output approximation by reducing parameters and enhancing accuracy. Direction-dependent scaling, shaping, and rotation optimize trend sensing for radial basis function networks.

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

Modeling of Cooperative Robotic Systems and Predictive Control Applied to Biped Robots and UAV-UGV Docking with Task Prioritization.

Sensors (Basel, Switzerland)·2024
Same author

The Use of Telemedicine to Improve Hypertension in an Urban Primary Care Clinic and Predictors of Improved Blood Pressure.

Journal of primary care & community health·2023
Same author

Reproducibility of an artificial intelligence optical coherence tomography software for tissue characterization: Implications for the design of longitudinal studies.

Cardiovascular revascularization medicine : including molecular interventions·2023
Same author

Multiple Dipole Source Position and Orientation Estimation Using Non-Invasive EEG-like Signals.

Sensors (Basel, Switzerland)·2023
Same author

Stochastic learning and extremal-field map based autonomous guidance of low-thrust spacecraft.

Scientific reports·2022
Same author

Optimization of Initial Drug Distribution in Spherical Capsules for Personalized Release.

Pharmaceutical research·2022

Area of Science:

  • Machine Learning
  • Artificial Intelligence
  • Numerical Analysis

Background:

  • Radial basis function (RBF) networks are powerful tools for function approximation.
  • Efficient parameterization and accurate trend sensing are crucial for RBF network performance.
  • Existing methods often require a large number of basis functions, impacting computational efficiency.

Purpose of the Study:

  • To introduce direction-dependent scaling, shaping, and rotation for Gaussian basis functions.
  • To develop novel formulations for minimal parameterization of general RBFs.
  • To present a directed graph-based algorithm for intelligent learning and adaptation in RBF networks.

Main Methods:

  • Incorporation of direction-dependent transformations (scaling, shaping, rotation) on Gaussian basis functions.

Related Experiment Videos

  • Development of alternate formulations for minimal parameterization of RBFs.
  • Introduction of a directed graph algorithm for adaptive parameter learning.
  • A parameter estimation algorithm using multiple data windows for initial parameter estimates.
  • Modification of the minimal resource allocating network (MRAN) for evaluation.

    • Main Results:

    • Shaping and rotation of RBFs reduce the number of function units needed for data approximation.
    • Improved accuracy is achieved with reduced parameter representations.
    • The directed graph algorithm facilitates intelligent, direction-based learning and adaptation.
    • The proposed methods demonstrate competitive or superior performance compared to existing algorithms on benchmark examples.

    Conclusions:

    • Direction-dependent transformations offer a significant advantage in RBF-based function approximation.
    • The novel directed graph algorithm enhances the learning and adaptation capabilities of RBF networks.
    • The presented approach achieves efficient and accurate input-output approximation with minimal parameters.