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

Toward an optimal PRNN-based nonlinear predictor.

D P Mandic1, J A Chambers

  • 1School of Information Systems, University of East Anglia, Norwich NR4 7TJ, UK.

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

Quaternion Common Spatial Patterns.

IEEE transactions on neural systems and rehabilitation engineering : a publication of the IEEE Engineering in Medicine and Biology Society·2016
Same author

A comprehensive study of the delay vector variance method for quantification of nonlinearity in dynamical systems.

Royal Society open science·2016
Same author

Dynamic response mitigation of floating wind turbine platforms using tuned liquid column dampers.

Philosophical transactions. Series A, Mathematical, physical, and engineering sciences·2015
Same author

Dynamic response signatures of a scaled model platform for floating wind turbines in an ocean wave basin.

Philosophical transactions. Series A, Mathematical, physical, and engineering sciences·2015
Same author

A regulatory phenotype associated with the en-am 1 mutant of Neurospora crassa.

Current genetics·2013
Same author

Trans-nuclear action of the nit-2 regulatory gene product and study of two additional nitrogen control genes in Neurospora crassa.

Current genetics·2013
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

We introduce an optimized parameter selection for pipelined recurrent neural networks (PRNNs) for nonlinear signal prediction. This method enhances prediction gain and overcomes vanishing gradient issues, outperforming existing predictors.

Area of Science:

  • Machine Learning
  • Signal Processing
  • Artificial Intelligence

Background:

  • Pipelined Recurrent Neural Networks (PRNNs) show promise for nonlinear and nonstationary signal prediction.
  • Existing PRNN training algorithms do not fully leverage inherent architectural features.
  • The problem of vanishing gradients in Recurrent Neural Networks (RNNs) hinders prediction accuracy.

Purpose of the Study:

  • To develop an approach for optimal parameter selection in PRNNs.
  • To investigate the role of nesting in PRNN architecture for signal prediction.
  • To enhance the prediction gain (PG) and computational efficiency of PRNNs.

Main Methods:

  • Analyzing the inherent nesting feature of the PRNN architecture.
  • Determining the optimal number of nested modules for specific prediction tasks.

Related Experiment Videos

  • Modifying the forgetting factor in the PRNN cost function based on nesting properties.

    • Main Results:

    • Nesting acts as a contractive function, allowing the forgetting factor to exceed unity, functioning as an emphasis factor.
    • This emphasis factor compensates for reduced contributions from distant modules, mitigating vanishing gradients.
    • The proposed PRNN parameter selection method significantly improves prediction gain compared to LMS, RLS, and prior PRNN schemes.

    Conclusions:

    • The developed parameter selection approach optimizes PRNN performance for nonlinear and nonstationary signal prediction.
    • PRNNs, when optimally configured, effectively address vanishing gradient issues.
    • The optimized PRNN achieves superior prediction accuracy without increased computational complexity.