Optimal Frequency Band Selection Based on Filter Banks and Wavelet Packet Decomposition in Multi-class Brain-Computer Interfaces

摘要

In multi-class BCI systems, the selection of frequency band is important in order to yield the system performance. Hence, it is necessary to select optimal frequency range for EEG feature extraction. For this end, feature selection method is used to find optimal frequency range. In general, discriminative features are often extracted from optimal frequency bands. Therefore, feature selection technique is capable of selecting optimal frequency signals. In this study, Fisher score was proved to be an efficient feature selection criterion.