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

Cascade associative memory storing hierarchically correlated patterns with various correlations.

M Hirahara1, N Oka, T Kindo

  • 1Matsushita Research Institute Tokyo, Inc., Kawasaki, Japan. mhira@mrit.mei.co.jp

Neural Networks : the Official Journal of the International Neural Network Society
|August 10, 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

Artificial intelligence in oncological positron emission tomography: advancing image analysis and interpretation.

Clinical radiology·2025
Same author

Anti-allergic effect of the Src family kinase inhibitor saracatinib.

Die Pharmazie·2020
Same author

A mild hand cleanser, alkyl ether sulphate supplemented with alkyl ether carboxylic acid and alkyl glucoside, improves eczema on the hand and prevents the growth of Staphylococcus aureus on the skin surface.

International journal of cosmetic science·2016
Same author

Discovering metabolic indices for early detection of squash (Cucurbita maxima) storage quality using GC-MS-based metabolite profiling.

Food chemistry·2015
Same author

Excitation Energy-Transfer Dynamics of Brown Algal Photosynthetic Antennas.

The journal of physical chemistry letters·2015
Same author

Search for bosonic superweakly interacting massive dark matter particles with the XMASS-I detector.

Physical review letters·2014

Conventional models for pattern storage assume uniform correlations, but real-world data shows varied correlations. The new CASM3 model effectively stores hierarchically correlated patterns with diverse similarity levels.

Area of Science:

  • Cognitive Science
  • Computational Neuroscience
  • Pattern Recognition

Background:

  • Conventional models for storing hierarchically correlated patterns assume uniform correlations between ancestor and descendant patterns.
  • This uniformity assumption is often unnatural, as object encoding typically preserves similarity, leading to varied correlation distances.
  • Consequently, conventional models become inapplicable when correlations are non-uniform.

Purpose of the Study:

  • To propose a novel model, CASM3, for storing hierarchically correlated patterns with varying correlation levels.
  • To address the limitations of conventional models that fail with non-uniform correlations.
  • To enhance the storage and retrieval of complex, similarity-preserving data patterns.

Main Methods:

Related Experiment Videos

  • Developed the CASM3 model, which accommodates variable correlations between hierarchical pattern levels.
  • Implemented a mechanism where critical load levels vary with descendants and increase with correlation.
  • Analyzed memory destruction patterns under increasing load levels, observing a descending order based on correlation.

    • Main Results:

    • CASM3 successfully stores hierarchically correlated patterns with diverse correlation distances.
    • Critical load levels in CASM3 are dynamically adjusted based on descendant correlations.
    • Memory destruction occurs in a predictable order, from highest to lowest correlation, as load increases.
    • The size of the basins of attraction is dependent on the correlation range, expanding with lower correlation levels.

    Conclusions:

    • The CASM3 model provides a more biologically and computationally plausible approach to storing hierarchically correlated patterns.
    • Variable correlations, rather than uniform ones, are crucial for accurately representing similarity-based encoding.
    • CASM3 offers improved robustness and capacity for memory storage in systems with complex, non-uniform relationships.