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

Smooth function approximation using neural networks.

Silvia Ferrari1, Robert F Stengel

  • 1Department of Mechanical Engineering and Materials Science, Duke University, Durham, NC 27708, USA. Sferrari@duke.edu

IEEE Transactions on Neural Networks
|March 1, 2005
PubMed
Summary

This study introduces an algebraic method for training feedforward neural networks to approximate complex functions. This approach offers faster training and improved generalization for neurocontrollers compared to traditional methods.

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

Extracorporeal Cytokine Hemadsorption for the Treatment of Severe Steroid Refractory Cytokine Release Syndrome Secondary to T-Cell Immunotherapy for B-Cell Malignancies: A Case Series.

Hematological oncology·2026
Same author

Social inclusion inequalities between individuals with mental disorders and the general population: a systematic review and meta-analysis.

Epidemiology and psychiatric sciences·2026
Same author

Clinical, economic, and organisational impact of the novel AAP 2022 neonatal hyperbilirubinemia guidelines in an Italian third-level neonatology unit.

European journal of pediatrics·2026
Same author

"Okay, So It's Not Me"-The <i>Extra-Fatigue</i> of Formal and Informal Reporting of Sexual Harassment in Academia.

International journal of environmental research and public health·2026
Same author

Case Report: Hermansky-Pudlak syndrome type 1 with pulmonary involvement in a 62-year-old Caucasian woman with rheumatoid arthritis.

Frontiers in immunology·2026
Same author

Cannabidiol for psychotic disorders: A meta-analysis of randomized controlled trials.

Journal of psychopharmacology (Oxford, England)·2026

Area of Science:

  • Artificial Intelligence
  • Machine Learning
  • Neural Networks
  • Numerical Analysis

Background:

  • Multidimensional nonlinear functions are challenging to represent accurately.
  • Traditional neural network training often relies on iterative optimization, which can be slow and prone to local minima.
  • Approximating functions with gradient information can improve network performance.

Purpose of the Study:

  • To present an algebraic approach for representing multidimensional nonlinear functions using feedforward neural networks.
  • To implement this approach for approximating smooth batch data, including input, output, and gradient information.
  • To investigate the training process and approximation properties through linear algebra.

Main Methods:

Related Experiment Videos

  • Representing nonlinear functions using feedforward neural networks.
  • Associating training data with network parameters via nonlinear weight equations.
  • Treating the cascade structure of weight equations as linear systems.
  • Developing four algorithms for exact or approximate matching of input-output and gradient-based training sets.

    • Main Results:

    • The algebraic approach allows for the investigation of neural network training and approximation properties via linear algebra.
    • Four distinct algorithms were developed for data matching.
    • Application to neurocontroller design demonstrated faster execution speeds.
    • The algebraic training method exhibited better generalization properties than contemporary optimization techniques.

    Conclusions:

    • An algebraic framework provides a novel perspective for understanding and training feedforward neural networks.
    • This method simplifies the analysis of complex function approximation.
    • The developed algorithms offer efficient and effective solutions for neurocontroller design, outperforming existing methods in speed and generalization.