Emotion classification from EEG signals using wearable sensors: pilot test
The objective of this work is to present a procedure for the classification of basic emotions based on the analysis of EEG signals (electroencephalogram). For this case, 25 subjects were stimulated, of whom 17 were men and 9 women between 20 and 35 years of age. The stimulus to induce positive, negative and neutral emotions with a certain level of excitation (activation) was a set of video clips previously evaluated. The processed and analyzed signals belong to the gamma and beta frequency bands of the F3, F4, P7, P8, T7, T8, O1 and O2 electrodes. The characteristic variables with the best result are the entropy of each band of each electrode. The cross validation algorithms are applied and later the main component analysis algorithm. Finally, four classifier algorithms are used: classifier trees, Support- Vector-Machine (SVM), Linear-Discriminant-Analysis (LDA) and k-Nearest-Neighbors (KNN). The results confirm that by carrying out the proposed procedure, the EEG signals contain enough information to allow the recognition of basic emotions.