Summary
International Symposium on Nonlinear Theory and its Applications
2005
Session Number:3-4-4
Session:
Number:3-4-4-1
Bootstrap Estimates for Nonlinear Predictors
Daisuke Haraki, Tomoya Suzuki, Tohru Ikeguchi,
pp.646-649
Publication Date:2005/10/18
Online ISSN:2188-5079
DOI:10.34385/proc.40.3-4-4-1
PDF download (363.6KB)
Summary:
Estimating Jacobian matrices of observed data generated by a nonlinear dynamical system is one of the important steps for nonlinear prediction. The Jacobian matrices are estimated by using local information about divergences of nearby trajectories. Although the basic algorithm for estimating the Jacobian matrices generally works well, it often fails for noisy data. In this paper, we proposed a new scheme to select a better near neighbor set for more accurate estimation of the Jacobian matrix: making a bootstrap subset of nearest neighbors. As a result, our method much improves nonlinear predictability not only for mathematical models with observational noise but also for real time serie