EEG feature extraction and classification using data dimension reduction

Analysis of biological signal plays a very important role in Brain Computer Interface (BCI). Particularly, with electroencephalogram (EEG), we can know the intension or mental state of human. To recognize those features, various parametric feature extraction methods such as central frequency, relative percent spectral energy band (RPEB), etc. is needed. In this paper, we propose an EEG signal classifier which handles time-domain EEG signal as a feature vector and reduces data dimension to create lower dimension features using in the classifier. We believe that the proposed method gives more reliable results than existing ones.