Emotion Recognition with the Feature extracted from brain Networks

Abstract

Emotion plays a crucial role in humans' daily life, which affects the decision and communication of human. Moreover, the effective recognition of emotion is essential to establish the affective Human-Computer Interaction (aHCI) systems. In this work, we mainly focus on feature extraction from the brain networks constructed with EEG to perform the emotion recognition. The analysis based on the public emotion dataset MAHNOB-HCI reveals that the proposed approach could achieved 100.00%, 99.95% and 99.99% for Negative-Neutral, Negative-Positive, and Positive-Neutral paired emotion states, respectively. Compared with the previous work for MAHNOB-HCI dataset, the proposed approach achieved the better classification results, and the experiment results have indicated that the feature extracted from brain networks is promising for the emotion classification.