Abstract

Info

Author(s):: WU Yanze1; WANG Hailing1; GAO Yufei2; 1. School of Electrical and Electronic Engineering, Shanghai University of Engineering Science, Shanghai 201600, China 2. School of Cyberspace Security, Zhengzhou University, Zhengzhou 450002, China

Keywords:: Keywords: electroencephalogram emotion recognition brain network model phase locking value graph theory minimum spanning tree machine learning

Abstract:: Abstract: The emotional brain networks at different time periods are constructed using phase locking values for exploring the stability of emotion-related brain network patterns at different time periods and an emotion recognition framework based on the second-order features of brain network is proposed. The results show that the highest accuracy are 79.17% and 82.92% in cross-subject study and single-subject study, and there is no significant difference in the recognition results of the 3 time periods in ANOVA analysis, which proves that the emotion recognition framework proposed in the study is stable. In different time periods, the emotions of the same category have the same brain network connection pattern and minimum spanning tree structure, indicating that the same emotion has a stable brain network pattern in different time periods. Using brain network features for emotion recognition is stable and reliable, which provides a new approach for emotion recognition in human-computer interaction.