EEG Signal Classification for Brain Computer Interface using SVM for Channel Selection

摘要

Brain-computer interface (BCI) is a emerging research area enabling the translation of motor imagery related brain signals into computer understandable signals by capturing the signal via Electroencephalogram or Electrocorticogram, processing the signal and classifying the motor imagery action. The BCI system detecting and identifying such motor signals can be used for numerous applications including controlling a computer or an electric wheel chair. In this paper the BCI paradigm is tested using Fast Hartley Transform for measuring energy and our proposed support vector machine algorithm to select the appropriate EEG channels required to solve the classification problem. We validate our hypotheses by applying this procedure to BCI Competition dataset IVA, a publicly available EEG repository. The evaluations of preprocessed signals showed that the extracted features were interpretable and can lead to high classification accuracy using the boosted tree classification algorithm.