Analysis of Multivariate Nonstationary Time Series Using the Localized Fourier Library
by Hernando Ombao, PhD (Department of Statistics, University of California at Irvine)
Furness Lecture Theatre 2
In this talk, we describe a set of methods for analyzing non-stationary time series using the SLEX library [(S)mooth (L)ocalized complex (EX)ponentials] as the primary tool. The SLEX library is a collection of bases; each of which consists of localized Fourier waveforms. In the problem of signal representation and spectral estimation we select, from the set of bases in the SLEX library, the one that best represents the underlying process that generated the data. Moreover, when discriminating nonstationary time series, we select the basis that gives the maximal separation between these classes. We illustrate the methods by analyzing multichannel electroencephalograms recorded during an epileptic seizure and during a visual-motor experiment.