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

Extended least squares based algorithm for training feedforward networks.

J F Yam1, T S Chow

  • 1City Univ. of Hong Kong, Kowloon.

IEEE Transactions on Neural Networks
|January 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

Filariasis in axillary lymph node.

Tropical biomedicine·2021
Same author

Identification and in vitro antimicrobial susceptibility of Brucella species isolated from human brucellosis.

International journal of microbiology·2014
Same author

A neural-based crowd estimation by hybrid global learning algorithm.

IEEE transactions on systems, man, and cybernetics. Part B, Cybernetics : a publication of the IEEE Systems, Man, and Cybernetics Society·2008
Same author

Blind equalization of a noisy channel by linear neural network.

IEEE transactions on neural networks·2008
Same author

Shape recovery from shading by a new neural-based reflectance model.

IEEE transactions on neural networks·2008
Same author

Learning parametric specular reflectance model by radial basis function network.

IEEE transactions on neural networks·2008
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 novel algorithm enhances feedforward network training by combining least squares and gradient descent methods. This approach achieves faster convergence and significantly reduces computational load compared to existing algorithms.

Area of Science:

  • Artificial Intelligence
  • Machine Learning
  • Neural Networks

Background:

  • Feedforward networks are widely used in machine learning.
  • Training feedforward networks can be computationally intensive.
  • Existing algorithms like Levenberg-Marquardt and backpropagation have limitations.

Purpose of the Study:

  • To propose an extended least squares-based algorithm for training feedforward networks.
  • To improve convergence speed and reduce computational cost.
  • To address the stalling problem in pure least squares algorithms.

Main Methods:

  • The algorithm first uses a least squares method to evaluate weights between the last hidden and output layers.
  • Modified gradient descent algorithms are then employed to evaluate weights between input and hidden layers.

Related Experiment Videos

  • This hybrid approach combines the strengths of both least squares and gradient descent.

    • Main Results:

    • The proposed algorithm eliminates the stalling problem associated with pure least squares methods.
    • It maintains fast convergence characteristics.
    • Significantly fewer floating-point operations (FLOPS) are required for convergence compared to the Levenberg-Marquardt algorithm (0.221%-16.0%).
    • The number of floating-point operations per iteration is slightly higher than standard backpropagation (1.517-3.521 times).

    Conclusions:

    • The extended least squares-based algorithm offers an efficient alternative for training feedforward networks.
    • It provides a favorable balance between convergence speed and computational efficiency.
    • This method presents a promising approach for reducing training time and resource requirements in deep learning applications.