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

Approximation of continuous and discontinuous mappings by a growing neural RBF-based algorithm.

A Esposito1, M Marinaro, D Oricchio

  • 1International Institute for Advanced Scientific Studies, Salerno, Italy.

Neural Networks : the Official Journal of the International Neural Network Society
|September 15, 2000
PubMed
Summary

This study introduces a novel neural network using Radial Basis Functions (RBF) for approximating continuous and discontinuous data. The network demonstrates improved approximation accuracy for complex mappings, offering enhanced performance in machine learning applications.

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

Application of the WHO Workload Indicators of Staffing Need (WISN) method to estimate workforce requirements in Italian Local Health Unit Prevention Departments.

BMC health services research·2026
Same author

A prospective study with Comprehensive Geriatric Assessment (CGA) before curative treatment in older patients with locally advanced head and neck squamous cell carcinoma (LA-HNSCC).

Oral oncology·2026
Same author

Off-label drug use in oncology: integrating molecular and genetic analysis in an academic center's real-world approach.

ESMO real world data and digital oncology·2026
Same author

Correlation between treatments and outcomes of patients with EGFR-mutated non-small-cell lung cancer that transitioned into small-cell lung cancer: an international retrospective study.

ESMO open·2025
Same author

miR-210 overexpression increases pressure overload-induced cardiac fibrosis.

Non-coding RNA research·2025
Same author

On a novel gradient flow structure for the aggregation equation.

Calculus of variations and partial differential equations·2024

Area of Science:

  • Machine Learning
  • Artificial Intelligence
  • Computational Neuroscience

Background:

  • Traditional neural networks struggle with approximating complex, discontinuous functions.
  • Radial Basis Functions (RBFs) offer a localized approach to function approximation.
  • Efficient learning strategies are crucial for optimizing neural network performance.

Purpose of the Study:

  • To develop and evaluate a novel neural network architecture for approximating continuous and discontinuous mappings.
  • To integrate an evolutionary optimization strategy for learning RBF variances.
  • To implement an incremental learning strategy for enhanced computational efficiency and performance.

Main Methods:

  • Utilized Radial Basis Functions (RBFs) as activation functions for hidden nodes.

Related Experiment Videos

  • Employed an evolutionary optimization strategy to learn the variances of RBFs.
  • Implemented a novel incremental learning strategy for selective network growth and local effect management.
  • Avoided the need for high-order derivatives in the learning process.

    • Main Results:

    • The developed neural network demonstrated superior performance in approximating continuous mappings compared to existing methods.
    • The network effectively handled discontinuous mappings with a finite number of discontinuities.
    • The incremental learning strategy resulted in significant computational time savings.
    • The approach showed improved approximation accuracy and efficiency.

    Conclusions:

    • The proposed RBF neural network offers a powerful and efficient solution for approximating complex continuous and discontinuous functions.
    • The integration of evolutionary optimization and incremental learning enhances network performance and computational efficiency.
    • This method provides a competitive alternative to existing approaches for challenging approximation tasks in machine learning.