Emotion recognition based on CNN-SVM from EEG signals

Emotion recognition is an important research directions in the field of artificial intelligence. In this paper, we adopted deep learning methods to try to improve the performance of emotion recognition. First, we designed an experiment to induce three different emotions by video footage and collected corresponding EEG signals. Then, CNN-SVM algorithms were adopted to classify positive, neutral and negative emotions. Finally, the recognition performance of CNN-SVM for three-class emotion recognition was verified. It is found that CNN-SVM achieves a highest recognition accuracy of 99.98% and an average recognition accuracy of 98.53%, which is better than CNN alone.