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

Intensity encoding in unsupervised neural nets.

Alan M. Parkinson1, Dawood Y. Parpia

  • 1School of Information Systems, Curtin University of Technology, 6001, Perth, Australia

Neural Networks : the Official Journal of the International Neural Network Society
|March 29, 2003
PubMed
Summary

Unsupervised neural networks lose signal intensity data due to input normalization. A novel encoding method preserves this information, enabling topological maps that maintain signal relationships across varying intensities and frequencies.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same journal

Injecting reasoning into vision-language models via weight-decomposed merging.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

A Non-equilibrium thermodynamic framework for neural networks: A principled correspondence and parameter dynamics.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

Dual-pathway mask ranking guided selective fine-tuning for backdoor purification.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

Discriminative transfer feature learning for unsupervised domain adaptation.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

ARetinex-Net: Low-light image enhancement with adaptive retinex model and global illumination representation.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

A Multi-Task Learning Framework with Physics Embedded for Signal Reconstruction and State Prediction in Underground Multi-Robot Systems.

Neural networks : the official journal of the International Neural Network Society·2026

Area of Science:

  • Artificial Intelligence
  • Computational Neuroscience
  • Signal Processing

Background:

  • Unsupervised neural networks often normalize input vectors, which can discard crucial signal intensity information.
  • Existing methods to retain intensity data, such as adding root-mean-square channels, are insufficient when fixed-length input vectors are required.

Purpose of the Study:

  • To address the information loss in normalized input vectors for unsupervised neural networks.
  • To propose and validate a new input vector encoding method that preserves signal intensity information.
  • To demonstrate the efficacy of this method using Kohonen Nets for topological mapping.

Main Methods:

  • Algebraic analysis to demonstrate information loss with standard normalization.
  • Development of a novel input vector encoding strategy by splitting channels.

Related Experiment Videos

  • Experimental validation using synthetic and cochlear model-based data with Kohonen Nets.

    • Main Results:

    • The proposed encoding method successfully retains signal intensity information within fixed-length input vectors.
    • Kohonen Nets trained with the new method form topological maps preserving relationships between signals of varying intensities and frequencies.
    • The network demonstrated appropriate responses to previously unencountered signals.

    Conclusions:

    • A novel input vector encoding method overcomes information loss in unsupervised neural networks.
    • This approach enables robust topological mapping of signals with diverse intensity and frequency characteristics.
    • The method shows promise for applications in signal processing and pattern recognition.