Mental and Emotional Recognition of College Students Based on Brain Signal Features and Data Mining

Nowadays, people pay more and more attention to the psychological situation of college students. Using data mining technology to model and analyze the collected psychological data of college students is a research hotspot in psychology and computer science. In addition, the essence of human emotional change is the higher nervous activity in the cerebral cortex. Electroencephalography (EEG) has become an important feature signal for emotion recognition because of its high time resolution and portability and practicality. Therefore, to solve the problem that the accuracy and generalization of the existing research models are not ideal, a method of college students’ psychological emotion recognition based on EEG signal features and data mining is proposed. Firstly, a feature selection method based on sparse learning is used to find out a few features from the high-dimensional feature space that contribute greatly to the reconstruction of category information so as to quickly acquire a few key emotion-related features. Then, the entropy-weighted clustering algorithm is combined with sparse learning feature selection, and the local structure of heterogeneous data is divided. Experimental results show that, compared with traditional methods, the proposed method has stronger applicability and higher accuracy of five categories of emotions, which provides a valuable reference for the evaluation of depression and anxiety of college students based on brain signal characteristics.