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

Neural classifiers using one-time updating.

K I Diamantaras1, M G Strintzis

  • 1Department of Applied Informatics, University of Macedonia, 540 06 Thessaloniki, Greece.

IEEE Transactions on Neural Networks
|February 7, 2008
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

High-speed multidimensional convolution.

IEEE transactions on pattern analysis and machine intelligence·2011
Same author

Optimal construction of subband coders using Lloyd-Max quantizers.

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

Use of nonlinear principal component analysis and vector quantization for image coding.

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

Optimal transform coding in the presence of quantization noise.

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

Optimal biorthogonal wavelet decomposition of wire-frame meshes using box splines, and its application to the hierarchical coding of 3-D surfaces.

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

Lossless coding of multichannel signals using optimal vector hierarchical decomposition.

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

This study introduces a novel theory for detecting linear nonseparability in pattern sets using a recursive computation procedure. The algorithm identifies separable subsets and offers a unique neural network implementation with committed synaptic weights, mimicking biological neural networks.

Area of Science:

  • Machine Learning
  • Artificial Intelligence
  • Computational Neuroscience

Background:

  • Linear threshold elements (LTEs), or perceptrons, are linear classifiers with limitations when faced with linearly nonseparable input patterns.
  • Sequential presentation of patterns poses challenges for traditional linear classification methods.

Purpose of the Study:

  • To develop a theory for the early detection of linear nonseparability in sequential pattern sets.
  • To create a recursive computation procedure for identifying and managing nonseparable patterns.
  • To propose a neural network implementation with unique weight dynamics and a learning procedure for convex class separation.

Main Methods:

  • Derivation of a theory for detecting linear nonseparability based on the solution cone in weight space.

Related Experiment Videos

  • Development of a recursive computation procedure for identifying the solution cone.
  • Identification and skipping of separability-violating patterns to obtain a separable subset.
  • Design of a neural network model with committed synaptic weights.
  • Combination of neural models to develop a convex class separation learning procedure.

    • Main Results:

    • A precise theory for detecting linear nonseparability as it emerges in sequential pattern sets.
    • A recursive algorithm enabling immediate detection of nonseparability.
    • Extraction of a totally separable subset from the original pattern set, along with its solution cone.
    • A neural network implementation with committed synaptic weights, akin to biological neural networks.
    • A learning procedure capable of separating convex classes by combining multiple neural models.

    Conclusions:

    • The proposed theory and algorithm effectively detect and manage linear nonseparability in pattern sets.
    • The developed neural network model offers a biologically plausible approach with committed synaptic weights.
    • The combined learning procedure provides a robust method for separating convex classes, advancing artificial neural network capabilities.