Analysis of EMG Based Emotion Recognition for Multiple People and Emotions

摘要

The human-machine interaction has limitations in recognizing emotions. Electromyogram (EMG) based emotion recognition is prominent. The lack of online availability of people’s EMG signal databank constraints in research works. This research generates an EMG signal databank of nine emotions for eleven people. Long and Short Term Memory classifier (LSTM) is used for EMG-based emotion recognition. LSTM classifier achieved a 51.77% improvement in classification accuracy than Least-Square Support Vector Machine (LSSVM) classifier.