EEG feature extraction algorithm based on CSP and R-CSP

摘要

Based on the traditional Common Spatial Pattern (CSP) algorithm, a new regularized public space pattern (RCSP) algorithm is proposed to process small sample electroencephalogram (EEG) data. R-CSP changes the values of the two regularization parameters so that the data of the experimenter has certain relevance and thus the method can reduce the error by individual differences. The method has been tested on public data sets. The results show that the classification results of R-CSP algorithm are better than the results of traditional CSP algorithm, and the effectiveness of the regularized CSP algorithm in the extraction of small sample EEG data is verified.