Classification of Emotional Arousal During Multimedia Exposure

Abstract

In the study of emotion recognition, relatively few efforts have been made to compare classification results across different emotion induction methods. In this study, we attempt to classify emotional arousal using physiological signals collected across three stimulus types -- music, videos, and games. Subjects were exposed to relaxing and exciting music and videos and then asked to play Tetris and Minesweeper. Data from GSR, ECG, EOG, EEG, and PPG signals were analyzed using machine learning algorithms. We were able to successfully detect emotion arousal over a set of contiguous multimedia activities. Furthermore, we found that the patterns of physiological response to each multimedia stimuli are varying enough, that we can guess the stimulus type just by looking at the biosignals.