Feature Extraction and Subset Selection for Classifying Single-Trial ECoG during Motor Imagery

The electrocorticogram (ECoG) recorded from subdural electrodes is a kind of BCI signal source that has the potential to achieve good classification results. The feature extraction and its subset selection are crucial for increasing classification accuracy rate. This paper proposes a new algorithm for classifying single-trial ECoG during motor imagery. The nonlinear regressive coefficients between signals on 10 leads are extracted in two frequency bands 0-3 Hz and 8-30 Hz as classification features. A genetic algorithm is used for the selection of the optimal feature subset and a support vector machine for their evaluation. The generalization error of 7% is achieved on data set I of BCI Competition III