A Novel Neural Approach For Classification of EEG Signals for Brain-Computer Interface
کد مقاله : 1190-CFIS (R1)
میلاد شوریابی *1، علی فروتن نیا2، علیرضا روحانی منش3، مهدیه قاسمی4
1مهندسی پزشکی-فنی مهندسی-دانشگاه سمنان-سمنان-ایران
2مهندسی پزشکی-فنی مهندسی-دانشگاه فردوسی مشهد-مشهد-خراسان رضوی-ایران
3مهندسی برق-فنی مهندسی-دانشگاه نیشابور-خراسان رضوی-ایران
4مهندسی برق-فنی مهندسی-دانشگاه نیشابور-نیشابور-خراسان رضوی-ایران
چکیده مقاله:
Nowadays, much research is done on Brain-Computer Interface (BCI). This interface communicates with the outside world using brain signals. Electroencephalography (EEG) proves that the largest measurement of brain activity is in BCI design. In this paper, 13 sets of EEG signals have been recorded from human participants in Neural Engineering Laboratory at University of Neyshabur, with the sampling frequency of 500 Hz, for three distinguishing tasks including up-down movement of left and right hands and eye blinking. After filtering and preprocessing, Fourier series is used for feature reduction and extraction. Then, Cascade-forward Neural Network is employed for pattern classification. The effect of different number of layers and neurons on classification accuracy is discussed. The experimental results demonstrate the proposed method is a simple and efficient approach for EEG-based BCI.
کلیدواژه ها:
Brain Computer Interface, Electroencephalography, Neural Network, Classification.
وضعیت : مقاله برای ارائه شفاهی پذیرفته شده است