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

Function approximation--a fast-convergence neural approach based on spectral analysis.

C Citterio1, A Pelagotti, V Piuri

  • 1Foster Wheeler Italiana S.p.A., 20094 Milano, Italy.

IEEE Transactions on Neural Networks
|February 7, 2008
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

Morphological and functional analysis of cryopreserved human sperm: comparison of different freezing protocols.

Cryo letters·2025
Same author

Real-life effectiveness and safety of baricitinib in patients with severe alopecia areata: A 24-week Italian study.

Journal of the European Academy of Dermatology and Venereology : JEADV·2024
Same author

Identification of a zoonotic genotype 3 hepatitis E subtype in wildlife in north-eastern Italy.

Infection, genetics and evolution : journal of molecular epidemiology and evolutionary genetics in infectious diseases·2019
Same author

Treatment of Vancouver B1, C periprosthetic hip fractures with periprosthetic polyaxial locking plate system: A 3-year follow-up.

Journal of biological regulators and homeostatic agents·2019
Same author

Characterization of Immune System Cell Subsets in Fixed Tissues from Alpine Chamois (Rupicapra rupicapra).

Journal of comparative pathology·2016
Same author

Efficacy, safety and acceptability of the new pen needle 33G × 4 mm. AGO 01 study.

Current medical research and opinion·2014
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

This study introduces a novel spectrum-based learning method for building neural networks. This approach efficiently approximates nonlinear functions, outperforming traditional techniques like backpropagation.

Area of Science:

  • Artificial Intelligence
  • Machine Learning
  • Computational Neuroscience

Background:

  • Neural networks are powerful tools for nonlinear function approximation.
  • Traditional training methods like backpropagation can be computationally intensive and slow.
  • Determining the optimal network architecture, particularly the number of hidden units, is often challenging.

Purpose of the Study:

  • To propose a constructive approach for building single-hidden-layer neural networks.
  • To introduce a spectrum-based learning procedure for efficient function approximation.
  • To enable automatic determination of the number of hidden units during training.

Main Methods:

  • Utilizing frequency domain analysis to guide network construction.
  • Implementing a spectrum-based learning procedure to minimize spectral differences between data and network estimates.

Related Experiment Videos

  • Employing an incremental network building process during training.

    • Main Results:

    • Achieved similar or superior nonlinear function approximation compared to traditional methods.
    • Demonstrated faster convergence times during the training process.
    • Successfully automated the determination of the required number of hidden units.

    Conclusions:

    • The proposed spectrum-based learning offers an efficient and effective alternative for training neural networks.
    • Frequency domain analysis provides a valuable framework for constructive neural network design.
    • This method simplifies network training by automatically managing architecture complexity.