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

Random noise effects in pulse-mode digital multilayer neural networks.

Y C Kim1, M A Shanblatt

  • 1Dept. of Electron. Eng., Chonnam Nat. Univ., Kwangju.

IEEE Transactions on Neural Networks
|January 1, 1995
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

Terahertz metamaterial-prism hybrid sensors for the detection of microorganisms.

Optics express·2025
Same author

Evaluating antimicrobial utilization in 20 Korean long-term care hospitals: a call to action for antimicrobial stewardship.

The Journal of hospital infection·2024
Same author

Clinical characteristics and management of orbital apex syndrome: a 10-year multicentre experience.

Rhinology·2024
Same author

Development of metastatic lung adenocarcinoma in a twenty-year-old skin graft site on the scalp: A case report.

Indian journal of dermatology, venereology and leprology·2023
Same author

Wells' syndrome following Oxford-AstraZeneca COVID-19 vaccination.

Journal of the European Academy of Dermatology and Venereology : JEADV·2022
Same author

Induction of the integrated stress response in the rat cornea.

Experimental eye research·2021
Same journal

Universal perceptron and DNA-like learning algorithm for binary neural networks: LSBF and PBF implementations.

IEEE transactions on neural networks·2013
Same journal

Guest editorial: special section on white box nonlinear prediction models.

IEEE transactions on neural networks·2011
Same journal

Data-based fault-tolerant control of high-speed trains with traction/braking notch nonlinearities and actuator failures.

IEEE transactions on neural networks·2011
Same journal

Guest editorial: special section on data-based control, modeling, and optimization.

IEEE transactions on neural networks·2011
Same journal

Neural network-based multiple robot simultaneous localization and mapping.

IEEE transactions on neural networks·2011
Same journal

Data-driven model-free adaptive control for a class of MIMO nonlinear discrete-time systems.

IEEE transactions on neural networks·2011
See all related articles

A novel digital multilayer neural network (DMNN) uses stochastic computing for efficient VLSI implementation. This approach offers greater accuracy than traditional methods, with a validated statistical model predicting performance.

Area of Science:

  • Computer Engineering
  • Artificial Intelligence
  • Digital Systems

Background:

  • Stochastic computing offers a promising alternative for implementing complex computational tasks, such as neural networks, using simple logic gates.
  • Digital Multilayer Neural Networks (DMNNs) are computationally intensive, posing challenges for efficient Very Large Scale Integration (VLSI) implementation.
  • Existing stochastic computing models often assume Bernoulli sequences, potentially underestimating achievable accuracy.

Purpose of the Study:

  • To implement a pulse-mode Digital Multilayer Neural Network (DMNN) utilizing stochastic computing principles.
  • To develop and validate a statistical model for analyzing noise and accuracy in stochastic neural network computations.
  • To compare the computational accuracy of the proposed DMNN with traditional deterministic approaches and existing stochastic models.

Related Experiment Videos

Main Methods:

  • Implemented a pulse-mode DMNN using basic logic gates and pseudorandom pulse sequences to replace algebraic neural operations.
  • Represented synaptic weights and neuron states as probabilities, estimated via average pulse occurrence rates.
  • Developed a statistical noise model based on hypergeometric distribution to predict computational accuracy and compared simulation results with model predictions.
  • Modeled DMNN feedforward architectures in VHDL and tested them using character recognition tasks.

Main Results:

  • The pulse-mode DMNN architecture is compact, flexible, and suitable for massively parallel VLSI implementation.
  • The developed statistical noise model accurately predicts the computational accuracy of the DMNN.
  • Experimental results demonstrate that the DMNN calculations are more accurate than anticipated under the common Bernoulli sequence assumption.

Conclusions:

  • The pulse-mode DMNN based on stochastic computing provides a viable and accurate approach for neural network implementation.
  • The developed statistical model offers a reliable method for performance analysis and prediction in stochastic neural networks.
  • This work advances the field of efficient neural network hardware design through innovative use of stochastic computing techniques.