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

First-order versus second-order single-layer recurrent neural networks.

M W Goudreau1, C L Giles, S T Chakradhar

  • 1Princeton Univ., NJ.

IEEE Transactions on Neural Networks
|January 1, 1994
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

Poster presentations at medical conferences: an effective way of disseminating research?

Clinical medicine (London, England)·2011
Same author

Optical computing: introduction by the feature editors.

Applied optics·2010
Same author

Learning, invariance, and generalization in high-order neural networks.

Applied optics·2010
Same author

Multiplexed coherent optical processor for calculating generalized moments.

Optics letters·2009
Same author

Neural networks and hybrid intelligent models: foundations, theory, and applications.

IEEE transactions on neural networks·2008
Same author

Attractive periodic sets in discrete-time recurrent networks (with emphasis on fixed-point stability and bifurcations in two-neuron networks).

Neural computation·2001
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

Second-order single-layer recurrent neural networks (SLRNNs) are more powerful than first-order SLRNNs. Augmenting first-order SLRNNs with feedforward layers and state-splitting enables finite-state recognizer implementation.

Area of Science:

  • Artificial Intelligence
  • Computational Neuroscience
  • Machine Learning

Background:

  • Recurrent neural networks (RNNs) are fundamental to sequence processing.
  • Single-layer recurrent neural networks (SLRNNs) offer a simpler architecture.
  • Understanding the representational limits of different RNN orders is crucial.

Purpose of the Study:

  • To compare the representational power of first-order and second-order SLRNNs with hard-limiting neurons.
  • To investigate methods for enhancing the capabilities of first-order SLRNNs.
  • To determine if first-order SLRNNs can implement finite-state recognizers.

Main Methods:

  • Theoretical analysis of first-order and second-order SLRNNs.
  • Examination of network architectures with hard-limiting activation functions.

Related Experiment Videos

  • Introduction and analysis of state-splitting technique in augmented SLRNNs.

    • Main Results:

    • Second-order SLRNNs possess strictly greater representational power than first-order SLRNNs.
    • Augmenting first-order SLRNNs with feedforward output layers allows for finite-state recognizer implementation.
    • The state-splitting technique is essential for enabling finite-state recognizer implementation in augmented first-order SLRNNs.

    Conclusions:

    • The order of recurrence in SLRNNs significantly impacts their representational capabilities.
    • Augmented first-order SLRNNs, with state-splitting, can efficiently implement any finite-state recognizer.
    • This research provides insights into the design and capabilities of recurrent neural networks for complex tasks.