Improved Filter Selection Method for Filter Bank Common Spatial Pattern for EEG-Based BCI Systems
Geun-Ho Park, Yu-Ri Lee, and Hyoung-Nam Kim
Pusan National University/Department of Electrical & Computer Engineering, Busan, Republic of Korea
Abstract—In this paper, we propose an improved filter selection method using Welch’s t-test based on discriminative filter bank common spatial pattern (DFBCSP). Existing DFBCSP used the Fisher ratio in order to find out discriminative filters. However, the Fisher ratio can be used to know only comparative value of distinguishability but may not become a meaningful criterion to reject null hypothesis. As a reasonable alternative, we have introduced the Welch’s t-test to find out not only contributory but also redundant filters used to classify features efficiently. Experimental results show that the classification accuracy increased by 1.28% on average when using the proposed filter selection method.
Index Terms—Brain-Computer Interface (BCI), Electroencephalography (EEG), Common Spatial Pattern (CSP), Discriminative Filter Bank Common Spatial Pattern (DFBCSP), Welch’s t-test
Cite: Geun-Ho Park, Yu-Ri Lee, and Hyoung-Nam Kim, "Improved Filter Selection Method for Filter Bank Common Spatial Pattern for EEG-Based BCI Systems," International Journal of Electronics and Electrical Engineering, Vol. 2, No. 2, pp. 101-105, June 2014. doi: 10.12720/ijeee.2.2.101-105
Index Terms—Brain-Computer Interface (BCI), Electroencephalography (EEG), Common Spatial Pattern (CSP), Discriminative Filter Bank Common Spatial Pattern (DFBCSP), Welch’s t-test
Cite: Geun-Ho Park, Yu-Ri Lee, and Hyoung-Nam Kim, "Improved Filter Selection Method for Filter Bank Common Spatial Pattern for EEG-Based BCI Systems," International Journal of Electronics and Electrical Engineering, Vol. 2, No. 2, pp. 101-105, June 2014. doi: 10.12720/ijeee.2.2.101-105
Array