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

An adaptive training method for optimal interpolative neural nets

T Z Liu1, C W Yen

  • 1Department of Mechanical Engineering, National Sun-Yat Sen University, Kaohsiung, Taiwan.

International Journal of Neural Systems
|April 1, 1997
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

Phylogeny and taxonomy of <i>Acer</i> powdery mildews, including genera <i>Sawadaea</i> and <i>Takamatsuella</i> (<i>Erysiphaceae, Ascomycota</i>).

Studies in mycology·2026
Same author

[A multicenter study on the effect of continued antiplatelet therapy on postprocedural bleeding after transrectal ultrasound-guided prostate biopsy].

Zhonghua yi xue za zhi·2025
Same author

Taxonomy and phylogeny of <i>Cortinarius</i> sect. <i>Anomali</i> in China.

Persoonia·2025
Same author

[Epidemiological characteristics of severe fever with thrombocytopenia syndrome in China, 2018-2021].

Zhonghua liu xing bing xue za zhi = Zhonghua liuxingbingxue zazhi·2024
Same author

[Minimally invasive treatment with function preservation for submucosal tumors in the gastric cardia].

Zhonghua wei chang wai ke za zhi = Chinese journal of gastrointestinal surgery·2020
Same author

Magnetotail reconnection onset caused by electron kinetics with a strong external driver.

Nature communications·2020

This study introduces improved methods for training optimal interpolative (OI) nets, enhancing performance and reducing complexity. The new approach offers greater stability and efficiency compared to existing least squares algorithms for machine learning.

Area of Science:

  • Machine Learning
  • Artificial Intelligence
  • Neural Networks

Background:

  • Conventional multilayered feedforward networks often rely on iterative gradient search for training.
  • Optimal interpolative (OI) nets offer a noniterative alternative using recursive least squares (RLS-OI).
  • RLS-OI utilizes subprototypes and prototypes from the training set to constrain and train the OI net.

Purpose of the Study:

  • To enhance the performance and reduce the complexity of OI net training.
  • To address critical issues in subprototype and prototype selection within the RLS-OI framework.
  • To develop adaptive and efficient methods for constructing OI nets.

Main Methods:

  • Proposed a novel, adaptive subprototype selection method focusing on poorly classified regions.

Related Experiment Videos

  • Introduced a new prototype selection criterion to minimize OI net complexity.
  • Dynamically increased subprototype and prototype numbers to evolve the OI net from scratch.

    • Main Results:

    • The proposed approach yielded smaller OI nets with equivalent training accuracy compared to standard RLS-OI.
    • Experimental results indicated reduced sensitivity to training set variations for the new method.
    • Simulations demonstrated improved classification performance and efficiency.

    Conclusions:

    • The novel subprototype and prototype selection strategies significantly improve OI net training.
    • The proposed methods offer a more robust and efficient alternative to existing RLS-OI algorithms.
    • This research contributes to the development of more effective and less complex neural network architectures.