EEG signal classification based on sparse representation in brain computer interface applications

Brain-Computer Interface (BCI) is a very essential and useful communication tool between the human brain and external devices. Effective and accurate classification of Electroencephalography (EEG) signals is important in performance of BCI systems. In this paper, a mental task classification approach based on sparse representation is proposed. A dictionary is used for classification, which is the combination of power spectral density calculated from EEG signal and common spatial pattern (CSP) algorithm. L1 minimization was used to classify EEG signals. Experimental results show that the proposed method provides higher classification performance compared to SVM and KNN classifiers. Based on the results average accuracy rates are as follows: 91.50%, 82.83%, 77.50% and 74%, for two, three, four and five classes, respectively.