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 Continuous-Time Asynchronous Boltzmann Machine.

Teruyuki Miyajima1, Masahiro Agu, Kazuo Yamanaka

  • 1Department of Systems Engineering and Ibaraki University, Japan

Neural Networks : the Official Journal of the International Neural Network Society
|August 1, 1997
PubMed
Summary

We introduce a novel asynchronous neural network model, mirroring the binary Hopfield model. This model achieves a steady-state fluctuation with probability distributions equivalent to serial Boltzmann machines, advancing neural network research.

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

Surgical Strategy of Intravenous Leiomyomatosis with Intracardiac Extension: A Case Report.

Annals of vascular diseases·2025
Same author

Long-term survival evaluation after transcatheter aortic valve implantation in patients with severe aortic valve stenosis: a retrospective cohort study.

Scientific reports·2025
Same author

Tritium transfer from seawater into marine organisms TFWT.

Radiation protection dosimetry·2024
Same author

[The Strategy for Aortic Root Pathology Due to Stanford Type A Acute Aortic Dissection].

Kyobu geka. The Japanese journal of thoracic surgery·2024
Same author

[The Long Outcome of the Left Atrial Appendage Preserving Maze Procedure].

Kyobu geka. The Japanese journal of thoracic surgery·2024
Same author

Interventricular septal dissection with perforations following takotsubo cardiomyopathy.

JTCVS techniques·2023

Area of Science:

  • Computational Neuroscience
  • Artificial Intelligence
  • Machine Learning

Background:

  • The binary Hopfield model is a foundational recurrent neural network architecture.
  • Understanding the dynamics and statistical properties of neural networks is crucial for advancing AI.
  • Serial Boltzmann machines offer a probabilistic framework for analyzing neural network states.

Purpose of the Study:

  • To propose a novel asynchronous neural network model.
  • To establish the equivalence in probability distribution between the proposed model and serial Boltzmann machines.
  • To explore the dynamic behavior of continuous-time neural networks.

Main Methods:

  • Developed an asynchronous neural network model with a structure analogous to the binary Hopfield model.

Related Experiment Videos

  • Modeled individual neuron behavior using continuous time dynamics.
  • Analyzed the steady-state behavior and probability distribution of the global network state.

    • Main Results:

    • The proposed asynchronous neural network model exhibits steady-state fluctuations.
    • The probability distribution of the global state in the proposed model is identical to that of a serial Boltzmann machine with equivalent synaptic weights.
    • Demonstrated a novel approach to modeling neural network dynamics.

    Conclusions:

    • The proposed asynchronous neural network model provides a new framework for studying neural computation.
    • Establishes a theoretical link between asynchronous continuous-time models and probabilistic models like Boltzmann machines.
    • Offers insights into the statistical properties and emergent behaviors of neural networks.