基于训练样本评估的CSP滤波器增量更新方法

由于脑电图(electro encephalo gram，EEG)能反映不同状态下大脑的思维活动，因此，基于EEG的运动想象识别已经成为一个新的研究热点。为了降低低质量样本对CSP(common spatial pattern)滤波器模型的组间传输性能的影响，提高正确率，提出了一种基于样本筛选的CSP滤波器增量更新方法。首先通过样本筛选的方法对EEG数据进行质量评估，然后剔除低识别率对应的单次训练数据，最后对优化后的样本所设计的CSP滤波器进行增量更新。实验室环境下，对EEG信号进行运动想象识别，其平均正确率达到80.92%，相比传统的CSP方法，五位受试者测试集的平均识别率分别提高了5.4%、5.6%、1.5%、8.6%和7.7%，实验结果验证了所提算法的有效性。

Incremental updating algorithm for CSP filter based on training sample evaluation

Electro encephalog ram(EEG) can reflect the thinking activity of the brain under different conditions, therefore, motor imagery recognition based on EEG has become a new research hot spot. To reduce the influence of low quality samples on the session-to-session transfer performance of CSP filter models and improve the recognition accuracy ratio, this paper proposed an incremental updating algorithm for CSP filter based on training sample evaluation. To start with, it used sample selection method to evaluate the quality of EEG data. Then it removed a set of training data corresponding to low recognition rate. Finally, it updated the CSP filter incrementally which designed by the optimized sample. In label environment, the motor imagery recognition of EEG signals reaches average accuracy of 80.92%. Compared with the traditional CSP method, the average recognition rate of the five subjects'' testing sets increases by 5.4%, 5.6%, 1.5%, 8.6%, and 7.7%, respectively. The experimental results verify the effectiveness of the proposed algorithm.