ub.xmlui.mirage2.page-structure.muninLogoub.xmlui.mirage2.page-structure.openResearchArchiveLogo
    • EnglishEnglish
    • norsknorsk
  • Velg spraakEnglish 
    • EnglishEnglish
    • norsknorsk
  • Administration/UB
View Item 
  •   Home
  • Fakultet for naturvitenskap og teknologi
  • Institutt for fysikk og teknologi
  • Artikler, rapporter og annet (fysikk og teknologi)
  • View Item
  •   Home
  • Fakultet for naturvitenskap og teknologi
  • Institutt for fysikk og teknologi
  • Artikler, rapporter og annet (fysikk og teknologi)
  • View Item
JavaScript is disabled for your browser. Some features of this site may not work without it.

Training Echo State Networks with Regularization Through Dimensionality Reduction

Permanent link
https://hdl.handle.net/10037/13086
DOI
https://doi.org/10.1007/s12559-017-9450-z
Thumbnail
View/Open
article.pdf (1.819Mb)
Submitted manuscript version (PDF)
Date
2017
Type
Journal article
Tidsskriftartikkel
Peer reviewed

Author
Løkse, Sigurd; Bianchi, Filippo Maria; Jenssen, Robert
Abstract
In this paper, we introduce a new framework to train a class of recurrent neural network, called Echo State Network, to predict real valued time-series and to provide a visualization of the modeled system dynamics. The method consists in projecting the output of the internal layer of the network on a lower dimensional space, before training the output layer to learn the target task. Notably, we enforce a regularization constraint that leads to better generalization capabilities. We evaluate the performances of our approach on several benchmark tests, using different techniques to train the readout of the network, achieving superior predictive performance when using the proposed framework. Finally, we provide an insight on the effectiveness of the implemented mechanics through a visualization of the trajectory in the phase space and relying on the methodologies of nonlinear time-series analysis. By applying our method on well-known chaotic systems, we provide evidence that the lower dimensional embedding retains the dynamical properties of the underlying system better than the full-dimensional internal states of the network.
Description
This is a pre-print of an article published in Cognitive Computation. The final authenticated version is available online at: https://doi.org/10.1007/s12559-017-9450-z.
Is part of
Løkse, S. (2020). Leveraging Kernels for Unsupervised Learning. (Doctoral thesis). https://hdl.handle.net/10037/19911.
Publisher
Springer Verlag (Germany)
Citation
Løkse, S., Bianchi, F.M. & Jenssen, R. (2017). Training Echo State Networks with Regularization Through Dimensionality Reduction. Cognitive Computation, 1-15. https://doi.org/10.1007/s12559-017-9450-z
Metadata
Show full item record
Collections
  • Artikler, rapporter og annet (fysikk og teknologi) [1057]

Browse

Browse all of MuninCommunities & CollectionsAuthor listTitlesBy Issue DateBrowse this CollectionAuthor listTitlesBy Issue Date
Login

Statistics

View Usage Statistics
UiT

Munin is powered by DSpace

UiT The Arctic University of Norway
The University Library
uit.no/ub - munin@ub.uit.no

Accessibility statement (Norwegian only)