Recognition of psychological emotion by EEG features

With the rapid development of emotion recognition technology, how to realize the naturalization and intellectualization of human–computer interaction, so that the human emotional state can be effectively recognized by the machine, and then get natural and harmonious emotional feedback results, has become the focus of research in the field of emotion recognition. The purpose of this study is to analyze the recognition of psychological emotions by electroencephalogram (EEG). In this study, Weibo, Fudan, and diary data sets were selected as data for psychological emotional research. The experimenter has mounted an electrode cap to test changes in the EEG signals when viewing various types of images, text and video. In this study, we extract and process EEG signals using wavelet transform and effectively identify and classify psychological emotions using support vector machines (SVM) and k-nearest neighbor algorithm. The results show that the average recognition rate of the improved grid PSVM method is 0.5% higher than that of the grid proximal support vector machine (PSVM) method; the average recognition rate of the improved grid RSVM method is 0.7% higher than that of the grid ranking support vector machines (RSVM) method; and the average recognition rate of the improved grid MKSVM method is 0.8% higher than that of the grid multi kernel support vector machine (MKSVM) method. It is concluded that the wavelet transform used in this study is more accurate in the extraction and processing of EEG features, and the method in this study is more accurate through the recognition of psychological emotions. It contributes to the recognition of psychological emotion by artificial intelligence.