Classification of mental tasks using support vector machine based on linear predictive coding and new mother wavelet transform

Abstract

The aims of Brain-Computer interfaces (BCI) research is helping paralyzed people communicating with others by using their electroencephalogram (EEG) signals. In this study, EEG signals from 5 mental tasks were recorded from 7 subjects and combinations of 2 different mental tasks were studied for each subject for one trial. The motivation for this work is using Linear predictive Coding (LPC) method to compress channels of EEG one channel. Eight features are employed for each signal of EEG using LPC 1st order followed by 3 level Discrete Wavelet Transform (DWT). New mother wavelet is used to be near the waveform of EEG signals. Statistical calculations are conducted for the 4 coefficients of DWT. Classification is conducted using support vector machine SVM. The classifier using SVM provided a high recognition rate reaching up to 100%, in some cases, and an average rate of about 85 %. The average specificity percent is 83.33 %. The average sensitivity percent is 86.66%.