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

A new neural network architecture with associative memory, pruning and order-sensitive learning.

M S Rao1, A K Pujari

  • 1Department of Computer & Information Sciences University of Hyderabad, India.

International Journal of Neural Systems
|December 10, 1999
PubMed
Summary

A novel neural network architecture functions as associative memory, incorporating pruning and order-sensitive learning. This advanced design ensures stable states without spurious ones, offering a capacity of 2n for enhanced data retrieval.

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

Statistical analysis of impact parameters of soft nose bullet fired on windowpane.

Forensic science international·2003
Same author

Characterization and differentiation of human embryonic stem cells.

Cloning and stem cells·2003
Same author

Hippocampal brain amines in methotrexate-induced learning and memory deficit.

Canadian journal of physiology and pharmacology·2002
Same author

Clitoria ternatea root extract enhances acetylcholine content in rat hippocampus.

Fitoterapia·2002
Same author

Effect of castration on dehydroepiandrosterone-induced hepatocarcinogenesis in male rats.

Anticancer research·2002
Same author

Clitoria ternatea (Linn) root extract treatment during growth spurt period enhances learning and memory in rats.

Indian journal of physiology and pharmacology·2002

Area of Science:

  • Artificial Intelligence
  • Computational Neuroscience
  • Machine Learning

Background:

  • Traditional neural networks often struggle with associative memory tasks, exhibiting limitations in handling order sensitivity and prone to spurious states.
  • Existing models lack efficient mechanisms for pruning and dynamic state convergence, hindering scalability and practical application.
  • The need for robust, high-capacity memory systems in AI drives the exploration of new network architectures.

Purpose of the Study:

  • To propose a novel neural network architecture functioning as an associative memory.
  • To integrate capabilities for pruning and order-sensitive learning within the network.
  • To demonstrate the network's effectiveness across diverse application domains.

Main Methods:

Related Experiment Videos

  • A composite neural network structure where each node is a Hopfield network.
  • Implementation of an order-sensitive learning technique within each Hopfield network node.
  • Design based on geometrical network structure and energy function for convergence to user-specified stable states without spurious states.
  • Incorporation of binary order pruning during associative memory retrieval.

    • Main Results:

    • The proposed network architecture successfully functions as an associative memory with pruning and order-sensitive learning capabilities.
    • The network demonstrates convergence to user-specified stable states, effectively avoiding spurious states.
    • Experimental validation across Library Database, Protein Structure Database, and Natural Language Understanding applications confirms its versatility.
    • The network achieves a storage capacity of 2n, where n is the number of basic nodes.

    Conclusions:

    • The novel composite Hopfield network architecture offers a significant advancement in associative memory systems.
    • Its inherent properties of order-sensitive learning, pruning, and spurious state avoidance make it highly suitable for complex data retrieval tasks.
    • The demonstrated success in diverse applications highlights its potential for real-world implementation in AI and data management.