摘要
不同的受试在进行运动想象时,脑电模式在频带分布上的差异较大,只有找到特定受试的有效特征,才能得到较好的实验效果。文中结合共同空间模型和粗糙集离散化算法的特征选择方法,来选取受试左右手运动想象的多频带脑电特征。与单频带特征相比,文中提出的方法提取的多频带脑电特征,能够有效的剔除了冗余特征量的干扰。实验结果表明(五位受试),文中提出的方法可以有效提高分类准确率。
The electroencephalogram features of Motor imagery vary considerably for different subjects. To achieve good performance of brain-computer interface, it is necessary to choose the features for specific subject. In this paper, we applied rough set based discretization algorithm to select feature from the feature vector of filter bank common spatial pattern. Comparing with single frequency band feature of common spatial pattern, the experimental results （Five subjects participated） show proposed method obtains better performance.
出处
《电子设计工程》
2014年第1期4-5,10,共3页
Electronic Design Engineering
关键词
脑—机接口
粗糙集理论
离散化算法
共同空间模型
brain-computer interface
rough set theory
discretization algorithm
common spatial pattern
