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

Comments on "Learning convergence in the cerebellar model articulation controller".

M Brown1, C J Harris

  • 1Dept. of Aeronaut. and Astronaut., Southampton Univ.

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

Case series: 2q33.1 microdeletion syndrome--further delineation of the phenotype.

Journal of medical genetics·2011
Same author

Clinical laboratory practices for the detection of rotavirus in England and Wales: can surveillance based on routine laboratory testing data be used to evaluate the impact of vaccination?

Euro surveillance : bulletin Europeen sur les maladies transmissibles = European communicable disease bulletin·2009
Same author

Fuzzy local linearization and local basis function expansion in nonlinear system modeling.

IEEE transactions on systems, man, and cybernetics. Part B, Cybernetics : a publication of the IEEE Systems, Man, and Cybernetics Society·2008
Same author

Linearization and state estimation of unknown discrete-time nonlinear dynamic systems using recurrent neurofuzzy networks.

IEEE transactions on systems, man, and cybernetics. Part B, Cybernetics : a publication of the IEEE Systems, Man, and Cybernetics Society·2008
Same author

Generalized neurofuzzy network modeling algorithms using Bézier-Bernstein polynomial functions and additive decomposition.

IEEE transactions on neural networks·2008
Same author

The relevance vector machine technique for channel equalization application.

IEEE transactions on neural networks·2008
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

The Cerebellar Model Articulation Controller (CMAC) neural network has limitations. Research shows the multivariate binary CMAC cannot learn all lookup tables, contradicting earlier theories on its learning capabilities.

Area of Science:

  • Computational Neuroscience
  • Artificial Intelligence
  • Machine Learning

Background:

  • The original Albus Cerebellar Model Articulation Controller (CMAC) theory proposed its capability to learn arbitrary multivariate lookup tables.
  • This theory suggested linear independence of basis functions and positive definiteness in its linear optimization process with sufficient data.
  • Prior work by Wong-Sideris (1992) supported these claims for the binary CMAC architecture.

Purpose of the Study:

  • To investigate the learning capabilities of the multivariate binary CMAC.
  • To challenge the established theory regarding the completeness and properties of the binary CMAC.
  • To provide a counterexample demonstrating the limitations of the original CMAC theory.

Main Methods:

  • Analysis of the multivariate binary CMAC architecture and its learning algorithms.

Related Experiment Videos

  • Mathematical proof and presentation of a 2D orthogonal function as a counterexample.
  • Demonstration of linear dependence in basis functions for both univariate and multivariate cases.

    • Main Results:

    • It has been proven that the multivariate binary CMAC is incapable of learning certain multivariate lookup tables.
    • The number of unlearnable functions grows exponentially with the generalization parameter.
    • Basis functions are always linearly dependent, rendering the optimization process positive semi-definite, leading to infinite optimal weight vectors.

    Conclusions:

    • The original theory on the arbitrary learning capacity of the binary CMAC is flawed.
    • The linear independence and positive definiteness assumptions do not hold for the multivariate binary CMAC.
    • These findings necessitate a re-evaluation of CMAC's theoretical underpinnings and practical applications in complex learning tasks.