kNN and SVM classification for EEG: a review

• Main Authors: Fuad, N., Sha'abani, M.N.A.H., Jamal, Norezmi, Ismail, M.F.
• Other Authors: Nasir, Ahmad Nor Kasruddin
• Format: Book Section
• Language: English
• Published: Springer Nature 2020
• Subjects: TK7800-8360 Electronics
• Online Access: http://eprints.uthm.edu.my/2872/1/kNN%20and%20SVM%20classification%20for%20eeg.pdf; http://eprints.uthm.edu.my/2872/
• Institution: Universiti Tun Hussein Onn Malaysia

This paper review the classification method of EEG signal based on k-nearest neighbor (kNN) and support vector machine (SVM) algorithm. For instance, a classifier learns an input features from a dataset using specific approach and tuning parameters, develop a classification model, and use the model to predict the corresponding class of new input in an unseen dataset. EEG signals contaminated with various noises and artefacts, non-stationary and poor in signal-to-noise ratio (SNR). Moreover, most EEG applications involve high dimensional feature vector. kNN and SVM were used in EEG classification and has been proven successfully in discriminating features in EEG dataset. However, different results were observed between different EEG applications. Hence, this paper reviews the used of kNN and SVM classifier on various EEG applications, identifying their advantages and disadvantages, and also their overall performances.