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

An experimental unification of reservoir computing methods.

D Verstraeten1, B Schrauwen, M D'Haene

  • 1Department of Electronics and Information Systems, Ghent University, Sint-Pietersnieuwstraat 41, 9000 Gent, Belgium. david.verstraeten@ugent.be

Neural Networks : the Official Journal of the International Neural Network Society
|May 23, 2007
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

Reservoir computing and extreme learning machines for non-linear time-series data analysis.

Neural networks : the official journal of the International Neural Network Society·2013
Same author

Locomotion without a brain: physical reservoir computing in tensegrity structures.

Artificial life·2012
Same author

Optoelectronic reservoir computing.

Scientific reports·2012
Same author

Information processing using a single dynamical node as complex system.

Nature communications·2011
Same author

Event detection and localization for small mobile robots using reservoir computing.

Neural networks : the official journal of the International Neural Network Society·2008
Same author

Mianserin and chronic pain: a double-blind placebo-controlled process and outcome study.

Acta psychiatrica Scandinavica·1993

This study compares three reservoir computing methods: Liquid State Machines (LSMs), Echo State Networks (ESNs), and Backpropagation Decorrelation (BPDC). A new Lyapunov exponent measure reveals optimal reservoir dynamics for improved performance.

Area of Science:

  • Computational Neuroscience
  • Machine Learning
  • Complex Systems

Background:

  • Recurrent neural networks (RNNs) are used as reservoirs in Liquid State Machines (LSMs), Echo State Networks (ESNs), and Backpropagation Decorrelation (BPDC).
  • Existing literature describes these techniques individually, lacking a comparative overview.
  • Understanding the relationship between reservoir parameters, network dynamics, and performance is crucial for advancing reservoir computing.

Purpose of the Study:

  • To provide a comprehensive comparison of LSMs, ESNs, and BPDC.
  • To investigate the impact of reservoir parameters on network dynamics, memory, node complexity, and benchmark performance.
  • To introduce and validate a novel measure for reservoir dynamics using Lyapunov exponents.

Main Methods:

Related Experiment Videos

  • Experimental comparison of LSM, ESN, and BPDC implementations across various benchmark tests.
  • Analysis of reservoir parameters including network dynamics, memory, and node complexity.
  • Introduction of a new reservoir dynamics measure based on Lyapunov exponents, sensitive to input-driven dynamics.

    • Main Results:

    • Experimental results demonstrate the performance variations among LSM, ESN, and BPDC under different conditions.
    • A strong correlation is found between reservoir parameters, network dynamics, and performance metrics.
    • The novel Lyapunov exponent measure indicates an optimal global scaling of the weight matrix, independent of traditional measures.

    Conclusions:

    • The study provides valuable insights into the comparative strengths and weaknesses of different reservoir computing approaches.
    • The new Lyapunov exponent-based measure offers a promising tool for optimizing reservoir computing systems.
    • The developed Reservoir Computing Toolbox facilitates further research and simulation of diverse reservoir computing architectures.