Summary
International Symposium on Nonlinear Theory and Its Applications
2016
Session Number:B2L-B
Session:
Number:B2L-B-4
Optimal Parameter Selection for Kernel Dynamic Mode Decomposition
Wataru Kurebayashi, Sho Shirasaka, Hiroya Nakao,
pp.-
Publication Date:2016/11/27
Online ISSN:2188-5079
DOI:10.34385/proc.48.B2L-B-4
PDF download (763.4KB)
Summary:
The dynamic mode decomposition is a recently proposed algorithm for the mode decomposition of multi-mode time series data based on the dynamical systems theory. The kernelization of this algorithm improves the estimation accuracy of dynamic modes, and therefore, facilitates wide-range applications of this method, while the optimal parameter selection for kernel dynamic mode decomposition remains still an open question. Here, we propose a formula for the patameter selection, and demonstrate the validity of our selection method.