Optimizing Common Spatial Pattern and feature extraction algorithm for Brain Computer Interface

Brain Computer Interface is the communication channel between the brain and the computer for recording of electrical activity along the scalp produced by the firing of neurons within the brain. The brain signals which are also known as Electroencephalography (EEG) can be used to direct and control some external activity. This work reports a methodology for acquisition and detection and of EEG signals, and extraction of useful information in order to differentiate the signals related to particular type of movement. A modified Common Spatial Pattern (CSP) algorithm has been used at preprocessing stage. Logarithmic transform along with the information theoretic feature extraction has also been used for feature extraction. KNN, SVM and Artificial Neural Networks are employed for classification. The proposed methodology is tested on publically available data sets and the results are found to be comparable with the published approaches.