DETERMINATION OF EMOTIONAL STATUS FROM EEG TIME SERIES BY USING EMD BASED LOCAL BINARY PATTERN METHOD

Authors : Ömer TÜRK

Pages : 313-321

Doi:10.36222/ejt.807971

View : 18 | Download : 19

Publication Date : 2020-12-30

Article Type : Research Paper

Abstract :Although determining emotional states from brain dynamics has been a subject that has been studied for a long time, the desired level has not been reached yet. In this study, Empirical mode decomposition insert ignore into journalissuearticles values(EMD); based Local Binary Pattern insert ignore into journalissuearticles values(LBP); method is proposed for emotional determination using insert ignore into journalissuearticles values(positive-neutral-negative); Electroencephalogram insert ignore into journalissuearticles values(EEG); signals. Thanks to this method, a hybrid structure was created in obtaining features from EEG signals. In the study, Seed EEG dataset containing 15 positive subjects and positive-neutral-negative emotional state is used. In the study, classification is utilized with the basis of individuals by using 27 EEG channels in the left hemisphere of each subject. Level 5 was separated by applying EMD to EEG segments containing three emotional states. Features were obtained from the Intrinsic mode function insert ignore into journalissuearticles values(IMF); using LBP method. These features are classified with k Nearest Neighbor insert ignore into journalissuearticles values(k-NN); and Artificial Neural Network insert ignore into journalissuearticles values(ANN);. The average classification accuracy for 15 participants was 83.77% using the k-NN classifier and 84.50% with the ANN classifier. In addition, the highest classification performance was found to be 96.75% with the k-NN classifier. The results obtained in the study support similar studies in the literature.
Keywords : EEG, Emotion, EMD, LBP, k NN, ANN