DISCRIMINATION AND CLUSTERING FOR MULTIVARIATE TIME-SERIES

Minimum discrimination information provides a useful generalization of likelihood methodology for classification and clustering of multivari ate time series. Discrimination between different classes of multivari ate time series that can be characterized by differing covariance or s pectral structures is of importance in applications occurring in the a nalysis of geophysical and medical time series data. For discriminatio n between such multivariate series, Kullback-Leibler discrimination in formation and the Chernoff information measure are developed for the m ultivariate non-Gaussian case. Asymptotic error rates and limiting dis tributions are given for a generalized spectral disparity measure that includes the foregoing criteria as special cases. Applications to pro blems of clustering and classifying earthquakes and mining explosions are given.