METHOD OF ANALYSING COMPOSITE COMMON SPATIAL PATTERN FOR BRAIN COMPUTER INTERFACE AND METHOD OF ANALYSING ELECTROENCEPHALOGRAM USING THE SAME

摘要

PURPOSE: A method of analyzing a composite common spatial pattern for a brain computer interface and a method of analyzing an electroencephalogram using the same are provided to improve the brain-wave classification performance for test target persons which are shortage of training data by using the metastasis between test target persons. CONSTITUTION: A composite covariance matrix is generated by using the composite covariance matrix of other test target persons(110). The composite covariance matrix corresponds to channels of time series data related to each classification of the test target persons. The eigen decomposition of the sum of the composite covariance matrixes is calculated(120). An eigenvector is extracted based on largest and smallest eigenvalues for the classification(130). Learning for classification is performed based on a label provided to the eigenvector and training data(140).