Feature selection on multi-physiological signals for emotion recognition

Recently, the most popular research in the field of emotion recognition on human-computer interaction is to recognize human's feeling using various physiological signals. In the psycho-physiological research, it is known that there is strong correlation between human emotion state and physiological reaction. In this study, seven kinds of emotion (happiness, sadness, anger, fear, disgust, surprise, stress) are evoked by audio-visual film clips as stimulation, and then autonomic nervous system responses as physiological signals are measured as the reaction of stimulation. In addition that, seven different emotions will be classified by the proposed classification methodology using physiological signals. We introduce a classification methodology on instance-based learning with feature selection that dwells upon the usage of biologically inspired optimization technique of Genetic Algorithms (GAs). In classification problems, it becomes important to carefully select prototypes and establish a subset of features in order to achieve a sound performance of a classifier. The study offers a complete algorithmic framework and demonstrates the effectiveness of the approach for the classification of seven emotions. Numerical experiments show that a suitable selection of prototypes and a substantial reduction of the feature space could be accomplished and the classifier formed in this manner is characterized by high classification accuracy for the seven emotions based on physiological signals.