Estimation of autocorrelation space for classification of bio-medical signals

Abstract

Classification of biomedical signals is a complex task, but the analysis is very useful in medical diagnosis. In this paper we estimate the autocorrelation matrix of some brain signal by embedding the autocorrelation cone using Linear Matrix Inequalities (LMI). The minimum sample window has been chosen for the improved computational complexity. The partitioning of the space has been carried out using support vector machines. This method has been tested on different EEG signals recorded on subjects performing a multiplication, thought for composition of a song. The base signature has been recorded while the subject apparently was not doing anything. © 2012 Springer-Verlag.