Multi-class Mental Task Classification Using Statistical Descriptors of EEG by KNN, SVM, Decision Trees, and Quadratic Discriminant Analysis Classifiers

Five mental tasks, baseline, multiplication, letter-composing, rotation, and counting, were studied. In our work, we evaluated the performance of six statistical features including the mean of raw signal, the standard deviation of raw signal, the mean of the absolute value of the first derivative of the signal, the mean of the absolute values of the first derivative of the normalized signal, the mean of the absolute values of the second derivative of the raw signal, and the mean of the absolute values of the second derivative of the normalized signal. K-nearest-neighbor, support vector machines, decision trees and quadratic discriminant analysis classifiers were designed. Mental task classification was investigated by using both hemispheres, right-hemisphere-only, and left-hemisphere-only. Accuracy rates were calculated between 85% and 100% for the KNN, SVM, and quadratic discriminant analysis classifiers.