Kernel Methods for Nonlinear Connectivity Detection

Autor(es)

Massaroppe, Lucas ^[1] ; Baccala, Luiz A. ^[2]

Afiliação do(s) autor(es)

^[1] Univ Sao Paulo, Dept Atmospher Sci, Inst Astron Geofis & Ciencias Atmosfer, BR-05508090 Sao Paulo - Brazil

^[2] Univ Sao Paulo, Escola Politecn, Dept Telecommun & Control Engn, BR-05508900 Sao Paulo - Brazil

• Tipo de documento: Artigo Científico
• Fonte: Entropy; v. 21, n. 6 JUN 2019.
• Citações Web of Science: 0

Resumo

In this paper, we show that the presence of nonlinear coupling between time series may be detected using kernel feature space <mml:semantics>F</mml:semantics> representations while dispensing with the need to go back to solve the pre-image problem to gauge model adequacy. This is done by showing that the kernelized auto/cross sequences in <mml:semantics>F</mml:semantics> can be computed from the model rather than from prediction residuals in the original data space <mml:semantics>X</mml:semantics>. Furthermore, this allows for reducing the connectivity inference problem to that of fitting a consistent linear model in <mml:semantics>F</mml:semantics> that works even in the case of nonlinear interactions in the <mml:semantics>X</mml:semantics>-space which ordinary linear models may fail to capture. We further illustrate the fact that the resulting <mml:semantics>F</mml:semantics>-space parameter asymptotics provide reliable means of space model diagnostics in this space, and provide straightforward Granger connectivity inference tools even for relatively short time series records as opposed to other kernel based methods available in the literature. (AU)

• Processo FAPESP: 15/50686-1 - Paleo-vínculos na evolução das monções e dinâmica
• Beneficiário: Pedro Leite da Silva Dias
• Modalidade de apoio: Auxílio à Pesquisa - Temático