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

Partially connected feedforward neural networks structured by input types.

Sanggil Kang1, Can Isik

  • 1Laboratory for Multimedia Computing, Communications, and Broadcasting, Information and Communications University, Daejeon, 305-714, South Korea. sang@icu.ac.kr

IEEE Transactions on Neural Networks
|March 1, 2005
PubMed
Summary

This study introduces a novel method for modeling partially connected feedforward neural networks (PCFNNs) by identifying input types (ITs). This approach enhances neural network structure and performance, demonstrated through experiments and blood pressure estimation.

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

Biased Pressure: Cyclic Reinforcement Learning Model for Intelligent Traffic Signal Control.

Sensors (Basel, Switzerland)·2022
Same author

Neural Network-Based Coronary Heart Disease Risk Prediction Using Feature Correlation Analysis

Journal of healthcare engineering·2017
Same author

Neural Network-Based Coronary Heart Disease Risk Prediction Using Feature Correlation Analysis.

Journal of healthcare engineering·2017
Same author

A reliable data delivery mechanism for grid power quality using neural networks in wireless sensor networks.

Sensors (Basel, Switzerland)·2011
Same author

A design of wireless sensor networks for a power quality monitoring system.

Sensors (Basel, Switzerland)·2011

Area of Science:

  • Artificial Intelligence
  • Machine Learning
  • Computational Neuroscience

Background:

  • Partially connected feedforward neural networks (PCFNNs) offer computational efficiency but require careful structural design.
  • Identifying input relationships (coupled vs. uncoupled) is crucial for optimizing PCFNNs.
  • Existing methods may not adequately capture the nuanced interactions between inputs in neural networks.

Purpose of the Study:

  • To propose a novel method for modeling PCFNNs based on the identification of input types (ITs).
  • To develop a technique for distinguishing between coupled and uncoupled inputs by analyzing sensitivity changes.
  • To demonstrate the effectiveness of the proposed method through experimental validation and a real-world application.

Main Methods:

Related Experiment Videos

  • Input sensitivity analysis was performed by amplifying input magnitudes to identify coupled and uncoupled inputs.
  • Uncoupled inputs showed sensitivity changes uncorrelated with other inputs.
  • Coupled inputs exhibited sensitivity changes correlated with variations in specific other inputs.
  • PCFNNs were structured based on identified ITs, with uncoupled inputs using independent neurons and coupled inputs sharing neurons.

    • Main Results:

    • The input sensitivity analysis successfully differentiated between coupled and uncoupled inputs.
    • The proposed method allowed for the tailored structuring of PCFNNs based on identified input relationships.
    • Experimental results and a blood pressure estimation case study validated the efficacy of the developed modeling approach.
    • The method demonstrated improved performance in modeling PCFNNs.

    Conclusions:

    • The proposed input type identification method provides a robust foundation for modeling PCFNNs.
    • Tailoring PCFNN architecture based on input coupling significantly enhances model performance.
    • This approach offers a valuable tool for designing efficient and effective neural networks for complex tasks.
    • The method's applicability is confirmed by its successful use in a physiological modeling example.