Chernihovskyi, Anton: Information-theoretic approach for the characterization of interactions in nonlinear dynamical systems. - Bonn, 2011. - Dissertation, Rheinische Friedrich-Wilhelms-Universität Bonn.
Online-Ausgabe in bonndoc: https://nbn-resolving.org/urn:nbn:de:hbz:5N-25132
@phdthesis{handle:20.500.11811/4975,
urn: https://nbn-resolving.org/urn:nbn:de:hbz:5N-25132,
author = {{Anton Chernihovskyi}},
title = {Information-theoretic approach for the characterization of interactions in nonlinear dynamical systems},
school = {Rheinische Friedrich-Wilhelms-Universität Bonn},
year = 2011,
month = may,

note = {Symbolic time series analysis provides us a solid and broadly used toolkit for the characterization of interactions between nonlinear dynamical systems. In this thesis, information-theoretic measures are evaluated with respect to their capability to characterize interactions between dynamical systems. We investigate several important limitations of these measures which may appear when experimental data exhibit strong correlations. It is demonstrated that a high degree of static and/or long-term temporal correlations can, in general, lead to the incorrect inference of directionality of interactions between underlying dynamical systems. In this thesis, we propose two complementary information-theoretic measures which can provide a better characterization of the directionality of interactions in cases where the influence of such correlations in data cannot be neglected.
First, the proposed information-theoretic measures are applied to characterize interactions between dynamical model systems with known equations of motion. Finally, they are applied to characterize interactions between multi-channel electroencephalographic recordings from epilepsy patients undergoing the presurgical diagnostics.},

url = {https://hdl.handle.net/20.500.11811/4975}
}

The following license files are associated with this item:

InCopyright