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

Neocognitron's parameter tuning by genetic algorithms.

D Shi1, C Dong, D S Yeung

  • 1Department of Electronics and Computer Science, University of Southampton, Highfield, UK. ds99r@ecs.soton.ac.uk

International Journal of Neural Systems
|January 29, 2000
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

First Measurement of Time-Dependent CP Violation in the Flavor-Changing Neutral-Current Decay B^{0}→K_{S}^{0}μ^{+}μ^{-}.

Physical review letters·2026
Same author

Measurement of the Top-Quark Production Cross Section and Charge Asymmetry at LHCb.

Physical review letters·2026
Same author

Searches for B^{0}→K^{+}π^{-}τ^{+}τ^{-} and B_{s}^{0}→K^{+}K^{-}τ^{+}τ^{-} Decays.

Physical review letters·2026
Same author

First Evidence of the B_{s}^{0}→K^{-}π^{+}γ Decay.

Physical review letters·2026
Same author

Precision Measurement of CP Violation and Branching Fractions in B^{±}→K_{S}^{0}h^{±} (h=π, K) Decays and Search for the Rare Decay B_{c}^{±}→K_{S}^{0}K^{±}.

Physical review letters·2026
Same author

First Observation of the B[over ¯]_{s}^{0}→Λ_{c}^{+}Λ[over ¯]_{c}^{-} Decay and Evidence for the B[over ¯]^{0}→Λ_{c}^{+}Λ[over ¯]_{c}^{-} Decay.

Physical review letters·2026
Same journal

Latent Space Projections and Atlases, a Cautionary Tale in Deep Neuroimaging using Autoencoders.

International journal of neural systems·2026
Same journal

Transformer-Based Anomaly Detection for Neurodegenerative Screening in MRI Images.

International journal of neural systems·2026
Same journal

Discrete Wavelet Convolution for Learnable Time-Frequency Representation with Application to Seizure Prediction.

International journal of neural systems·2026
Same journal

Automatic Seizure Detection using Hierarchical Spectral-Temporal Feature Learning with an Imbalance-Aware Transformer.

International journal of neural systems·2026
Same journal

Pyramid Vision Transformer-Enhanced Conformer Network for Epileptic Seizure Recognition Using MultiChannel EEG Signals.

International journal of neural systems·2026
Same journal

A Time-Frequency Decoupled Contrastive Learning Framework for Electroencephalography-Based Parkinson's Disease Diagnosis.

International journal of neural systems·2026
See all related articles

This study optimizes the Neocognitron, a pattern recognition system, by using Genetic Algorithms (GAs) to tune parameters and training patterns. Results show GAs improve Neocognitron performance, demonstrating sensitivity to training data and parameters.

Area of Science:

  • Artificial Intelligence
  • Computational Neuroscience
  • Machine Learning

Background:

  • Fukushima's Neocognitron recognizes distorted patterns but requires manual parameter tuning and empirical training data.
  • Supervised learning in Neocognitron is layer-by-layer, with parameters like selectivity and receptive fields set manually.
  • Existing methods for Neocognitron training are limited by manual parameter setting and empirical training pattern design.

Purpose of the Study:

  • To investigate the sensitivity analysis of Neocognitron.
  • To develop a Genetic Algorithm (GA)-based approach for optimizing Neocognitron parameters and training patterns.
  • To enhance Neocognitron's pattern recognition capabilities through automated tuning.

Main Methods:

  • Utilized Genetic Algorithms (GAs) to tune Neocognitron parameters and search for optimal training pattern sets.

Related Experiment Videos

  • Incorporated cooperative coevolution to handle the large search space of training patterns, overcoming limitations of traditional GAs.
  • Developed an effective fitness function for applying GA-based optimization to numeral recognition tasks.

    • Main Results:

    • Demonstrated that Neocognitron performance is sensitive to its training patterns, selectivity, and receptive fields.
    • Showed that performance does not necessarily increase monotonically with the number of training patterns.
    • Confirmed that the GA-based supervised learning approach can significantly improve Neocognitron's performance.

    Conclusions:

    • The correlation analysis between training patterns and Neocognitron performance is validated.
    • Tuning the number of planes in Neocognitron is equivalent to searching for effective training patterns.
    • GA-based supervised learning offers an effective method for enhancing Neocognitron's pattern recognition accuracy and robustness.