Mixed Emotion Recognition Based on EEG Signals

Advanced emotion monitoring and intervention systems are critical for human-machine cooperation to finish specific tasks in which recognizing mixed emotions is essential. This study is the first to propose intelligent recognition methods of mixed emotions based on EEG signals, thus endowing machines with higher emotional intelligence. We extracted differential entropy (DE), differential asymmetry (DASM), and rational asymmetry (RASM) as features. To better serve the system application and promotion, this study selected three lightweight classifiers, including Random Forest (RF), Probabilistic Neural Networks (PNN), and Generalized Regression Neural Networks (GRNN), all of which achieved good classification performance. The highest classification accuracy is 96.7%. The results of this study can be applied in the fields of mental state assessment, psychological assistance, and clinical psychotherapy.