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

A modified error backpropagation algorithm for complex-value neural networks.

Xiaoming Chen1, Zheng Tang, Catherine Variappan

  • 1Faculty of Engineering, Toyama University, 3190 Gofuku, Toyama-shi, Toyama 930-8555, Japan. xmchen1@hotmail.com

International Journal of Neural Systems
|December 31, 2005
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

A Non-Canonical Role of SMAD4 in Regulating 3D Genome Architecture to Inhibit Lung Squamous Cell Carcinoma Development.

Advanced science (Weinheim, Baden-Wurttemberg, Germany)·2026
Same author

Loss-induced spatial division of Smith-Purcell radiation from even-parity phase gradient metasurfaces.

Optics letters·2026
Same author

Enabling efficient electron injection in stretchable OLED.

Nature materials·2025
Same author

Diffraction regimes of phase gradient metasurfaces in background media with symmetry breaking.

Optics express·2025
Same author

Immune-compatible designs of semiconducting polymers for bioelectronics with suppressed foreign-body response.

Nature materials·2025
Same author

Knocking out artificially selected gene GmAOC4<sup>H8</sup> improves germination in soybean.

TAG. Theoretical and applied genetics. Theoretische und angewandte Genetik·2025

This study introduces a modified error function for complex-valued backpropagation, effectively preventing local minima and accelerating learning in telecommunications and signal processing applications.

Area of Science:

  • Artificial Intelligence
  • Signal Processing
  • Machine Learning

Background:

  • Complex-valued backpropagation is crucial for telecommunications, speech recognition, and image processing.
  • A significant challenge in this algorithm is the occurrence of local minima during the learning process.
  • This limitation hinders optimal model training and performance.

Purpose of the Study:

  • To address the local minima problem in complex-valued backpropagation.
  • To enhance the speed of the learning process.
  • To propose a novel modification to the conventional error function.

Main Methods:

  • A modified error function was developed by incorporating a term related to the hidden layer error.
  • This approach aims to guide the learning process away from suboptimal solutions.

Related Experiment Videos

  • Simulations were conducted to evaluate the algorithm's effectiveness.

    • Main Results:

    • The proposed modified error function successfully prevented the algorithm from getting stuck in local minima.
    • A notable increase in the learning speed was observed.
    • Simulation results validated the efficacy of the new approach.

    Conclusions:

    • The modified error function offers a robust solution to the local minima problem.
    • The algorithm demonstrates improved learning efficiency and speed.
    • This advancement has significant implications for complex-valued neural network applications.