2019 : Modifikasi Fitur dengan Differential Asymmetry untuk Meningkatkan Akurasi Klasifikasi EEG Motor Imagery

Mochamad Hariadi ST., M.Sc., Ph.D
Dr. Tri Arief Sardjono S.T, MT.


Abstract

Brain-Computer Interface (BCI) technology has enabled people with motor disabilities to interact with their environment. The electroencephalograph (EEG) signals related to a motor imagery movement were used as a control signal. In this paper, EEG motor imagery signals from the 2-class data have been processed into features and classified. The power and standard deviation of EEG signals, mean of absolute wavelet coefficients, and the average power of the wavelet coefficients were used as features. The purpose of this paper is to apply the differential asymmetry of these features as new features to improve the system accuracy. As a classifier, Support Vector Machine (SVM), k-Nearest Neighbor (k-NN), and Tree were used. The result shows that for dataset I the use of differential asymmetry as feature can increase the system accuracy up to 47.8%, from 52.20% to 100%, with Tree as a classifier. For dataset II, it can increase accuracy by 8.46%, from 54.42% to 62.48%.