A1004
Title: Linear discriminant analysis of high-dimensional time series
Authors: Danna Zhang - University of California, San Diego (United States) [presenting]
Abstract: Classification is one of the fundamental problems in time series analysis, where the goal is to assign a new observed series to one of multiple known classes. While the classification of low-dimensional time series has been well studied, the investigation of high-dimensional cases remains limited. Sparse linear discriminant analysis is applied to high-dimensional time series, and conditions for the consistency of the time series LDA rule are established for both Gaussian and non-Gaussian processes. Numerical studies and application to fMRI data are conducted to corroborate the results.
