Classification of multiclass eeg signal related to mental task using higuchi fractal dimension and 10-Statistic Parameters - Support Vector Machine

Nowadays, Not only the accuracy of a classification system but also a feature extraction method is an important matter in a Brain Computer Interface Application. In this paper, we investigated the multiclass classification of mental task using EEG signal. Higuchi Fractal Dimension and 10-Statistic Parameters were used as feature extraction method. The 10-statistic parameters are central tendency type that is, maximum value, minimum value, mean, standard deviation, median, mode, variance, first-quartile, third-quartile, interchange quartile. Multiclass Support Vector Machine with One-against-All strategy is applied to classify EEG signal related to the mental task. The result shows that the Multiclass SVM classifier with 1-against-All strategy using 10-Statistic Parameters has a higher accuracy when compared to Higuchi Fractal Dimension-SVM, Extreme Learning Machine, Back Propagation Neural Network, both of Support Vector Machine 1-versus-1 strategy and 1-versus-All strategy. The average accuracy ranging between 99.2% and 100% for 10-Statistic Parameters-SVM and HFD_SVM ranging from 60.22% to 91.91% were gained for five mental task classes.