Classification of imagined writing from EEG signals using autoregressive features

Imagined writing is one of the techniques that may improve writing disorder when brain is trained to perform the activity. The imagined writing activity embedded in EEG signal can be extracted and classified using Autoregressive model and Multi Layer Perceptron. This paper describes the classification of imagined writing letters from EEG signals using Multi Layer Perceptron with Autoregression model as feature extraction method. The optimum Autoregression model order was determined by examining the classification accuracy of Multi Layer Perceptron under various orders. The results showed that the best range of Autoregression order for classifying imagined letters from EEG signals is 15 to 20. With this range, the classification accuracy increases linearly.