Feature exaction and classification of attention related electroencephalographic signals based on sample entropy

Abstract

A method regarding the sample entropy (SampEn) as features is proposed to carry out the analysis and classification of attention related electroencephalographic(EEG) signals, and the support vector machine (SVM) algorithm is used as classifiers for classification, seven males (aged from 20 to 30) are recruited to perform three attention-related tasks, including attention, inattention, and relaxation states. The processing results demonstrate that the classification accuracy of the SampEn gets up to 85.5% for classifying the relation between attention and inattention, obviously much higher than that with frequency band power (77.9%). It indicates that the SampEn is more effective to extract the information attention-related in EEG to show the clinical application prospects in EEG biofeedback systems.