An Artificial Neural Network approach for electroencephalographic signal classification towards brain-computer interface implementation

Brain-Computer Interface (BCI) can be realized by translating user's thoughts into control commands to assist paralyzed persons to communicate and control electronic devices. In this work, Electroencephalographic (EEG) signals were recorded from four subjects while they perform different mental states. We present an Artificial-Neural-Network-based approach for the purpose of classifying Electroencephalographic signals into different mental states which are equivalent to different control commands for our BCI implementation. Inputs of the Artificial Neural Network are spectral features dimensionally reduced by Principal Component Analysis. Experimental results show that the proposed method outperforms other classifiers, i.e., K-Nearest Neighbor, Naive Bayesian, Support Vector Machine, and Linear Discriminant Analysis in our EEG dataset with highest classification results on dual and triple mental state problems of 95.36% and 76.84%, respectively.