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

Model-free distributed learning.

A Dembo1, T Kailath

  • 1Stanford Univ., CA.

IEEE Transactions on Neural Networks
|January 1, 1990
PubMed
Summary
This summary is machine-generated.

This study introduces model-free learning for quasi-static networks, enabling continuous weight adjustments via signal correlation. This distributed approach facilitates integrated, on-chip learning in large analog and optical systems.

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

Eigenstructure approach to directions-of-arrival estimation in IR detector arrays.

Applied optics·2010
Same author

Climate change and the integrity of science.

Science (New York, N.Y.)·2010
Same author

Depth of focus and the moment expansion.

Optics letters·2009
Same author

Estimation of multiple 2-D uniform motions by SLIDE: subspace-based line detection.

IEEE transactions on image processing : a publication of the IEEE Signal Processing Society·2008
Same author

A randomized trial of standard versus partially hyperfractionated radiation with or without concurrent 5-fluorouracil in locally advanced cervical cancer.

Gynecologic oncology·1998
Same author

X-ray small angle scattering study of chromatin as a function of fiber length.

Molecular biology reports·1998
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

Area of Science:

  • Artificial Intelligence
  • Machine Learning
  • Network Science

Background:

  • Traditional network learning often requires detailed models and central control.
  • Variability and defects in large networks pose significant challenges for implementation.
  • Existing methods struggle with integrated, on-chip learning for analog and optical systems.

Purpose of the Study:

  • To present a novel model-free learning approach for synchronous and asynchronous quasi-static networks.
  • To enable distributed, on-chip learning capabilities in large-scale analog and optical networks.
  • To develop a mechanism invariant to network structure variations and implementation defects.

Main Methods:

  • Continuous perturbation of network weights using noise sources or orthogonal signals.

Related Experiment Videos

  • Measurement and correlation of a time-varying performance index with perturbation signals.
  • Weight updates determined by the correlation output, forming a local and distributed learning rule.

    • Main Results:

    • Demonstrated a learning mechanism that is invariant to detailed network structure, reducing variability.
    • Showcased a completely distributed mechanism requiring minimal global signals and no central control.
    • Enabled integrated, on-chip learning suitable for large analog and optical networks.

    Conclusions:

    • The proposed model-free learning method offers a robust and scalable solution for network adaptation.
    • This approach overcomes limitations of traditional methods by enabling decentralized and integrated learning.
    • The technique is particularly advantageous for large analog and optical networks where central control is impractical.