Emotion recognition employing ECG and GSR signals as markers of ANS

Change in GSR and ECG signals are expression of Autonomic Nervous System response to change in environment and physiological system. In this study, we evaluate the effectiveness of GSR and ECG in emotional state recognition. We took into account three mutually exclusive emotional stateshappy, sad and neutral and three different approaches to recognize these emotional states. The results indicate that to classify opposite valence emotion (happy and sad) GSR solely is capable of recognizing emotions with enhanced accuracy with a minimum accuracy of 99.44%. However, in classifying emotionally active state (happy or sad) in reference to neutral state, combined feature set of both ECG and GSR is more effective, especially the combination of PSD features and SVM classifier, providing a high accuracy for classifying all three emotional states.