Summary
the 2014 International Symposium on Nonlinear Theory and its Applications
2014
Session Number:C1L-D
Session:
Number:C1L-D3
Inferring functional connectivities of networks with discrete and continuous observables
Ryosuke Hosaka,
pp.446-449
Publication Date:2014/9/14
Online ISSN:2188-5079
DOI:10.34385/proc.46.C1L-D3
PDF download (342.4KB)
Summary:
We here introduce estimating methods of functional connectivities of the networks, under the condition that the observable variables are composed of discrete and continuous ones. The causality estimation in the frequency domain is based on the estimation of the cross-spectrum density matrix. So, the main problem is how do we estimate the cross-specral density matrix. The cross-spectral density matrix is estimated through the multi- variate auto-regressive models for the continuous variables. In the current study, the discrete observables were assumed to be event time-series (e.g. timing of the earth quake and firing timing of the nerve cells). For the discrete variables, the cross-spectral density matrix can be estimated by the Fourier transformation and multitaper methods. In the current study we also consider the situation that the observed variables are composed of the discrete and continuous variables. For example, recording of the brain signals often provides the spike event time-series and the EEG or LFP signals. The spike signal is discrete and the LFP is continuous signals. For such cases, the event-triggered continuous averaging is useful method for the cross-spectral density matrix estimation. The current manuscript provides the outline of the estimation methods for the functional connectivity.