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

Pattern recognition in the neocognitron is improved by neuronal adaptation

A van Ooyen1, B Nienhuis

  • 1Netherlands Institute for Brain Research, Amsterdam.

Biological Cybernetics
|January 1, 1993
PubMed
Summary

Adding neural adaptation to Fukushima's neocognitron significantly enhances pattern discrimination. This adaptation allows circuits to preferentially extract key distinguishing features, improving network performance.

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

The effect of limited sensory information on exoskeleton performance in people with complete spinal cord injury.

IEEE ... International Conference on Rehabilitation Robotics : [proceedings]·2022
Same author

Immediate after-effects of robot-assisted gait with pelvic support or pelvic constraint on overground walking in healthy subjects.

Journal of neuroengineering and rehabilitation·2019
Same author

Estimating severity of sideways fall using a generic multi linear regression model based on kinematic input variables.

Journal of biomechanics·2017
Same author

Totally asymmetric simple exclusion process simulations of molecular motor transport on random networks with asymmetric exit rates.

Physical review. E, Statistical, nonlinear, and soft matter physics·2015
Same author

Incorporating in vivo fall assessments in the simulation of femoral fractures with finite element models.

Medical engineering & physics·2015
Same author

Shear banding of colloidal glasses: observation of a dynamic first-order transition.

Physical review letters·2014

Area of Science:

  • Artificial Intelligence
  • Computational Neuroscience
  • Machine Learning

Background:

  • The neocognitron is a hierarchical, multilayered neural network architecture.
  • Original neocognitron models learn features shared across patterns, potentially limiting discrimination.
  • Neural adaptation, a decrease in neuron activity upon repeated stimulation, is a biological phenomenon.

Purpose of the Study:

  • To investigate the impact of incorporating neural adaptation into the neocognitron architecture.
  • To determine if adaptation enhances the pattern discriminatory power of the network.
  • To understand how adaptation influences the development of feature-extracting circuits.

Main Methods:

  • Modification of the neocognitron model to include adaptive neuron behavior.

Related Experiment Videos

  • Simulation and analysis of network performance with and without adaptation.
  • Comparison of feature extraction circuits developed in both network variants.

    • Main Results:

    • Equipping neocognitron neurons with adaptation markedly improved pattern discriminatory power.
    • Adaptive neocognitron preferentially developed circuits for extracting discriminating features.
    • The original neocognitron preferentially learned shared features, hindering discrimination.

    Conclusions:

    • Neural adaptation is a crucial mechanism for enhancing pattern discrimination in neocognitron networks.
    • Adaptive circuits facilitate the extraction of unique, pattern-specific features.
    • Incorporating biological neural phenomena like adaptation can significantly advance artificial neural network capabilities.