We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

physics

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Physics > Data Analysis, Statistics and Probability

Title: Stochastic approach for assessing the predictability of chaotic time series using reservoir computing

Abstract: The applicability of machine learning for predicting chaotic dynamics relies heavily upon the data used in the training stage. Chaotic time series obtained by numerically solving ordinary differential equations embed a complicated noise of the applied numerical scheme. Such a dependence of the solution on the numeric scheme leads to an inadequate representation of the real chaotic system. A stochastic approach for generating training times series and characterising their predictability is suggested to address this problem. The approach is applied for analysing two chaotic systems with known properties, Lorenz system and Anishchenko-Astakhov generator. Additionally, the approach is extended to critically assess a reservoir computing model used for chaotic time series prediction. Limitations of reservoir computing for surrogate modelling of chaotic systems are highlighted.
Comments: Paper published as part of the special topic on In Memory of Vadim S. Anishchenko. Dataset used for figures and some additional comments on the used numerical methods are openly available in this http URL
Subjects: Data Analysis, Statistics and Probability (physics.data-an)
MSC classes: 34K23, 34K50, 34F05, 60H10, 37H10,
ACM classes: G.1; I.6; J.2
Journal reference: Published in 'Chaos', Volume 31, 083105 (2021)
DOI: 10.1063/5.0058439
Cite as: arXiv:2110.05483 [physics.data-an]
  (or arXiv:2110.05483v1 [physics.data-an] for this version)

Submission history

From: Igor Khovanov A [view email]
[v1] Mon, 11 Oct 2021 08:34:21 GMT (1636kb)

Link back to: arXiv, form interface, contact.