Kernel-based relevant feature extraction to support Motor Imagery classification

摘要: Brain Computer Interface (BCI) has become one of the most interesting alternatives to support automatic systems able interpret brain functions. Recently, Motor Imagery (MI) paradigm is a widely topic interest as tool develop BCI-based systems. Here, we present relevant feature extraction methodology, termed MI discrimination using kernel relevance analysis (MIDKRA), classification in BCI For such purpose, similarity criterion rank contribution EEG features for classifying an employed. The introduced approach includes supervised information regarding find out set encoding discriminative information. We model recordings by considering both time and time-frequency representations. Moreover, k nearest-neighbor classifier carried validate proposed approach. Experimental results on two different databases, well-known public data Emotiv-based dataset built us, demonstrate that MIDKRA outperforms state art methods it suitable alternative straightforward