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

CNN universal machine as classificaton platform: an art-like clustering algorithm.

David Bálya1

  • 1Analogical & Neural Computing Laboratory, HAS Computer and Automation Research Institute, Kende u. 13, Budapest, H-1111, Hungary. balya@sztaki.hu

International Journal of Neural Systems
|March 20, 2004
PubMed
Summary

This study maps adaptive resonance theory (ART) to cellular neural/nonlinear network universal machines (CNN-UM) for robust feature vector classification. The developed analogic algorithm achieves high accuracy, enabling efficient real-time object recognition.

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

Genetic reactivation of cone photoreceptors restores visual responses in retinitis pigmentosa.

Science (New York, N.Y.)·2010
Same author

Genetically timed, activity-sensor and rainbow transsynaptic viral tools.

Nature methods·2009
Same author

Light-activated channels targeted to ON bipolar cells restore visual function in retinal degeneration.

Nature neuroscience·2008
See all related articles

Area of Science:

  • Computational neuroscience
  • Artificial intelligence
  • Machine learning

Background:

  • Fast and robust classification of feature vectors is essential for real-time systems.
  • Cellular neural/nonlinear network universal machines (CNN-UM) excel as feature detectors.
  • Post-processing of detected features is critical for object recognition.

Purpose of the Study:

  • To demonstrate the mapping of a robust classification scheme based on adaptive resonance theory (ART) to the CNN-UM.
  • To develop a generalizable analogic CNN algorithm for feature vector classification.
  • To implement and evaluate the performance of this algorithm on standard CNN-UM chips.

Main Methods:

  • Developed an analogic CNN algorithm based on ART principles.
  • Extended the algorithm for both unsupervised and supervised classification tasks.

Related Experiment Videos

  • Implemented the binary feature vector classification on existing CNN-UM hardware.

    • Main Results:

    • The analogic CNN algorithm effectively classifies extracted feature vectors.
    • The classification scheme retains the robust, plastic, and fault-tolerant properties of ART networks.
    • Experimental evaluation demonstrated promising performance, achieving 100% accuracy on the training set.

    Conclusions:

    • The proposed ART-based classification scheme is suitable for CNN-UM implementation.
    • This approach offers a robust and efficient solution for real-time feature vector classification.
    • The algorithm's flexibility supports tunable sensitivity and automatic class creation.