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

Using function approximation to analyze the sensitivity of MLP with antisymmetric squashing activation function.

D S Yeung1, Xuequan Sun

  • 1Dept. of Comput., Hong Kong Univ.

IEEE Transactions on Neural Networks
|February 5, 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

Assessing the interchangeability of keratometry measurements from four biometric devices in intraocular lens power calculations: insights into the predictive accuracy of five modern IOL formulas.

BMC ophthalmology·2025
Same author

Re: Cheng et al.: Conbercept versus laser for the treatment of infants with zone II retinopathy of prematurity (Ophthalmology. 2024;131:636-638).

Ophthalmology·2025
Same author

Comparative Analysis of Eighteen IOL Power Calculation Formulas Using a Modified Formula Performance Index Across Diverse Biometric Parameters.

American journal of ophthalmology·2025
Same author

A mosquito proboscis-inspired cambered microneedle patch for ophthalmic regional anaesthesia.

Journal of advanced research·2024
Same author

Machine Learning Tools to Assist the Synthesis of Antibacterial Carbon Dots.

International journal of nanomedicine·2024
Same author

Eye Transient Ischemia Attack Full Record.

Retina (Philadelphia, Pa.)·2024
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 generalized method for analyzing neural network sensitivity before design. It offers insights into network parameters, aiding in structure determination and weight range estimation for training.

Area of Science:

  • Artificial Intelligence
  • Computational Neuroscience
  • Machine Learning

Background:

  • Neural network sensitivity analysis is typically performed post-design and training.
  • Existing methods, like Piche's, have limitations on input and weight perturbations.

Purpose of the Study:

  • To generalize Piche's statistical method for multilayer perceptron (MLP) sensitivity analysis.
  • To derive a universal expression for MLP sensitivity applicable to antisymmetric squashing activation functions.
  • To provide a method for network design decisions based on sensitivity analysis.

Main Methods:

  • Derivation of a universal MLP sensitivity expression for antisymmetric squashing activation functions.
  • Removal of restrictions on input and output perturbations.

Related Experiment Videos

  • Experimental validation using a three-layer MLP with 30 nodes per layer.

    • Main Results:

    • Theoretical investigations closely align with experimental results.
    • Analysis reveals the impact of network design parameters (layers, neurons, activation function).
    • A novel network design method is proposed.

    Conclusions:

    • The generalized sensitivity analysis offers a critical approach to neural network design.
    • The method aids in determining optimal network structure and permissible weight ranges for training.
    • Findings provide valuable guidance for making informed network design decisions.