Performance Analysis of Multiclass Common Spatial Patterns in Brain-Computer Interface
Brain-Computer Interfacing (BCI) aims to assist, enhance, or repair human cognitive or sensory-motor functions. The classification of EEG signals plays a crucial role in BCI implementation. In this paper we have implemented a multi-class CSP Mutual Information Feature Selection (MIFS) algorithm to classify our EEG data for three class Motor Imagery BCI and have presented a comparative study of different classification algorithms including k-nearest neighbor (kNN) and Fuzzy kNN algorithm, linear discriminant analysis (LDA), Quadratic discriminant analysis (QDA), support vector machine (SVM), radial basis function (RBF) SVM and Naive Bayesian (NB) classifiers algorithms. It is observed that Fuzzy kNN and kNN algorithm provides the highest classification accuracy of 92.65% and 92.29% which surpasses the classification accuracy of the other algorithms.
KeywordsBrain-Computer Interfacing Electroencephalography Common Spatial Pattern Mutual Information Features Selection k-Nearest Neighbor Fuzzy k-Nearest Neighbor Linear Discriminant Analysis Quadratic Discriminant Analysis Support Vector Machine Nave-Bayesian
- 7.Wu, W., Gao, X., Gao, S.: One-versus-the-rest (OVR) algorithm: An extension of common spatial patterns (CSP) algorithm to multi-class case. In: 27th Annual International Conference of the Engineering in Medicine and Biology Society (IEEE-EMBS), pp. 2387–2390. IEEE Press, Shanghai (2006)Google Scholar
- 9.Lotte, F., Congedo, M., Lecuyer, A., Lamarche, F., Arnaldi, B.: A Review of Classification Algorithms for EEG-based Brain-Computer Interfaces. J. Neural Eng. 4, R1–R13 (2007)Google Scholar
- 10.Theodoridis, S., Koutroumbas, K.: Pattern Recognition. Academic Press (2006)Google Scholar