An innovative emotion assessment using physiological signals based on the combination mechanism

The main purpose of this paper is the assessment of emotions using Electroencephalogram (EEG) and peripheral physiological signals and improvement of recognition accuracy of emotional states using combination mechanism. In the rst step, according to the type of signals, eective features were extracted in the time and frequency domains; then, by using the Fisherʹs Linear Discriminant (FLD) method, the most eective features were selected. Based on these features, six classiers were used: Support Vector Machine (SVM), Nearest Mean (NM), K-Nearest Neighborhood (K-NN), 1-Nearest Neighborhood (1-NN), FLD, and Linear Discriminant Analysis (LDA). They classied emotions in two classes (low and high) through arousal, valence, and liking dimensions. The Leave-One-Out Cross-Validation (LOOCV) method has been implemented to evaluate the performance of classiers. To enhance the accuracy of classication, combination at feature and classier levels was performed. Via the concatenation method, combination at feature level was done. Then, by Majority voting, Fixed and Stacking algorithms, combination at classier level was implemented. Results showed that these classiers were selected properly and, thanks to them, good improvements were achieved compared with previous studies. Finally, by using combination methods, obtained recognition accuracy was much more reliable and combination at classier level resulted in signicant improvement.