Novel method for detection of idle state in brain-computer interface system
异步脑—机接口的空闲状态检测新方法*

Thie paper proposed a novel method for detecting idle state. With within-class scatter and the accuracy of the motor samples in training set as two indexes, the proposed method used the idea of ROC to determine the thresholds, and designed an optional 3-classifier. And made a fuzzy on the predicted value at the same time. Obtained a final test result with mean squared error of 0.278 7 on the BCI competition 2005 dataset IVc. The result validates the efficiency of the proposed method.