Classifying Motor Imagery EEG Signals by Iterative Channel Elimination according to Compound Weight

摘要

There often exist redundant channels in EEG signal collection which deteriorate the classification accuracy. In this paper, a classification method which can deal with redundant channels, as well as redundant CSP features, is presented for motor imagery task. Our method utilizes CSP filter and margin maximization with linear programming to update a compound weight that enables iterative channel elimination and the update of the following linear classification. Theoretical analysis and experimental results show the effectiveness of our method to solve redundancy of channels and CSP features simultaneously when classifying motor imagery EEG data.