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

Improved bidirectional retrieval of sparse patterns stored by Hebbian learning.

Friedrich T. Sommer1, Gunther Palm

  • 1Department of Neural Information Processing, University of Ulm, 89069, Ulm, Germany

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

This study introduces a new crosswise bidirectional (CB) retrieval algorithm for the Willshaw model, significantly reducing errors in hetero-associative memory tasks. The improved algorithm achieves high information efficiency, approaching theoretical limits for neural associative memories.

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

Comparing Surrogates to Evaluate Precisely Timed Higher-Order Spike Correlations.

eNeuro·2022
Same author

Artificial Development by Reinforcement Learning Can Benefit From Multiple Motivations.

Frontiers in robotics and AI·2021
Same author

Detection and evaluation of bursts in terms of novelty and surprise.

Mathematical biosciences and engineering : MBE·2019
Same author

Iterative Retrieval and Block Coding in Autoassociative and Heteroassociative Memory.

Neural computation·2019
Same author

Neural Information Processing in Cognition: We Start to Understand the Orchestra, but Where is the Conductor?

Frontiers in computational neuroscience·2016
Same author

Predicting Variabilities in Cardiac Gene Expression with a Boolean Network Incorporating Uncertainty.

PloS one·2015

Area of Science:

  • Computational Neuroscience
  • Artificial Intelligence
  • Information Theory

Background:

  • The Willshaw model, a neural associative memory (NAM), offers asymptotic efficiency but suffers from high retrieval errors in its finite form.
  • Iterative retrieval methods have been explored to enhance performance in auto-association tasks.

Purpose of the Study:

  • To define and determine the asymptotic bound of information efficiency for the bidirectional Willshaw model in hetero-association tasks.
  • To propose and analyze an efficient new bidirectional retrieval strategy for the finite Willshaw model.
  • To explore the implications of a sparse bidirectional associative memory (BAM) for applications and brain theory.

Main Methods:

  • Derivation of a combinatorial formula for faster numerical evaluation of the dendritic sum distribution.

Related Experiment Videos

  • Development and analysis of a novel crosswise bidirectional (CB) retrieval algorithm.
  • Simulation experiments to evaluate the performance of the CB retrieval strategy.

    • Main Results:

    • The proposed CB retrieval significantly reduces crosstalk error without complex learning rules or pattern augmentation.
    • Combinatorial analysis and simulations demonstrate that CB retrieval achieves high information efficiency, nearing the asymptotic bound.
    • The naive extension of the Willshaw model with bidirectional retrieval is shown to be unpromising.

    Conclusions:

    • The CB retrieval algorithm offers a significant improvement for sparse bidirectional associative memory, enabling efficient hetero-associative mapping and auto-associative completion.
    • The enhanced BAM model has potential applications in information retrieval, including data access, input segmentation, and relevance feedback.
    • Bidirectional associative memory models provide functional insights into reciprocal cortico-cortical pathways and Hebbian cell-assemblies.