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

Parameter incremental learning algorithm for neural networks.

Sheng Wan1, Larry E Banta

  • 1Mechanical and Aerospace Engineering Department, West Virginia University, Morgantown, WV 26505, USA. Sheng_Wan@yahoo.com

IEEE Transactions on Neural Networks
|November 30, 2006
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

Ginseng-mulberry (medicine-food homologous) pair mitigates cadmium-induced anxiety: a clinical proteomics-guided network pharmacology with rat validation.

Frontiers in psychiatry·2026
Same author

Time-updated explainable machine learning predicts short-term mortality in peritoneal dialysis patients.

Renal failure·2026
Same author

Rapid and Highly Selective Detection of the Tire Antioxidant 6PPD in Water Samples via Dual-Mode Fluorescent and Photoelectrochemical Sensing Based on Host-Guest Recognition.

Environmental science & technology·2026
Same author

Impaired Midkine/Treg Signaling Contributes to Pathogenesis of Preeclampsia.

Hypertension (Dallas, Tex. : 1979)·2026
Same author

Dynamic dose-time proteomic profiling of cadmium exposure in rat prefrontal cortex.

Neurotoxicology·2026
Same author

Prediction of Imminent Peritoneal Dialysis-Associated Peritonitis Using Time-Updated Electronic Health Records and Machine Learning: A Temporal Validation Study.

Journal of inflammation research·2026
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

A new parameter incremental learning (PIL) algorithm enhances neural network training. This novel approach adapts to new data while preserving past learning, outperforming standard methods in speed and accuracy.

Area of Science:

  • Artificial Intelligence
  • Machine Learning
  • Neural Networks

Background:

  • Neural network training algorithms often struggle to balance adaptation to new data with retention of previously learned information.
  • Existing online training methods like backpropagation can be inefficient or inaccurate when faced with evolving datasets.

Purpose of the Study:

  • To introduce a novel stochastic training algorithm for neural networks called parameter incremental learning (PIL).
  • To develop a PIL algorithm that adapts to new input-output patterns while preserving prior learning.
  • To evaluate the performance of the PIL algorithm against established methods.

Main Methods:

  • A general PIL algorithm for feedforward neural networks was formulated as an approximate solution to an optimization problem.

Related Experiment Videos

  • The optimization problem incorporated measures for both data adaptation and preservation of prior results.
  • Specific PIL algorithms were derived for the multilayer perceptron (MLP).

    • Main Results:

    • Numerical studies demonstrated that the PIL algorithm for MLP significantly outperformed the standard online backpropagation (BP) and stochastic diagonal Levenberg-Marquardt (SDLM) algorithms.
    • The PIL algorithm showed superior convergence speed and accuracy across three benchmark problems.
    • The PIL algorithm was found to be computationally simple and easy to use, similar to the BP algorithm.

    Conclusions:

    • The parameter incremental learning (PIL) algorithm offers a superior alternative to existing online training methods for neural networks.
    • PIL effectively balances adaptation and preservation, leading to enhanced performance in terms of speed and accuracy.
    • The algorithm's simplicity and effectiveness suggest broad applicability in scenarios where online BP is currently used.