Brain Computer Interface Signals Classification for Right and Left Hands Imagined Movements
Mohammad Rafi Barakzai1, Manjaiah D.H2
1Mohammad Rafi*, Computer Science, Mangalore University, Mangalore, Karnataka, India.
2Dr. Manjaiah D.H, Computer Science, Mangalore University, Mangalore, Karnataka, India.
Manuscript received on March 12, 2020. | Revised Manuscript received on March 25, 2020. | Manuscript published on March 30, 2020. | PP: 3268-3269 | Volume-8 Issue-6, March 2020. | Retrieval Number: F8035038620/2020©BEIESP | DOI: 10.35940/ijrte.F8035.038620
Open Access | Ethics and Policies | Cite | Mendeley
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Brain signals are complex and nonstationary, each signal represents an intendent behavior of a user. Brain Computer Interfaces (BCIs) are used to extract and translate these signals. Electroencephalography (EEG) is the common method used for this purpose. Identifying which signal represents which action is important. In this paper, we extracted spatial and spectral patterns for right and left hands imagined movements from EEG signals. We considered only C3 and C4 bipolar channels and band frequency (8-30 Hz) for both alpha and beta. The relevant features classified using Support Vector Machine (SVM) and 9 other common classifiers to compare and contrast classification performances. To ensure classification performance we calculated confusion matrix classes for all 9 subjects from II b dataset. Classification accuracy observed and recorded from all classifiers and 9 subjects. The highest classification accuracy scores for 3 subjects S4, S8 and S9 are (100 %, 89 %, and 71 %) and misclassification scores are (0 %, 11% and 29 %) respectively.
Keywords: BCI, EEG, Machine Learning, Support Vector Machine, Signal Processing
Scope of the Article: Machine Learning.