Computer classification of the EEG time series by Kullback information measure

Abstract
Kullback information plays an important role in measuring the discrepancy between 2 probability density functions. Kullback information was proven equivalent to the spectral error measure, which was developed to estimate the difference between 2 spectral densities. To divide the non-stationary electroencephalogram into stationary subsequences, segmentations of the data during sleep were performed using the spectral error measure. To classify the segmented electroencephalogram into one of the template pattern classes, Kullback information was computed. Kullback information was used as a measure of the distance between the template pattern and the segmented pattern. Human sleep stages determined by the segmentation and the classification of the electroencephalogram, were similar to those determined by a medical doctor.

This publication has 7 references indexed in Scilit: