Identification of motor imagery tasks through CC-LR algorithm in brain computer interface

Abstract

This study focuses on the identification of Motor Imagery (MI) tasks for the development of Brain Computer Interface (BCI) technologies combining Cross-Correlation and Logistic Regression (CC-LR) techniques. The proposed method is tested on two benchmark data sets, IVa and IVb of BCI Competition III, and the performance is evaluated through a 3-fold cross-validation procedure. The experimental outcomes are compared with two recently reported algorithms, R-Common Spatial Pattern (CSP) with aggregation and Clustering Technique (CT)-based Least Square Support Vector Machine (LS-SVM) and also other four algorithms using data set IVa. The results demonstrate that our proposed method results in an improvement of at least 3.47% compared with the existing methods tested.