Bagging RCSP脑电特征提取算法

发布时间：2018-03-14 05:14

本文选题：脑电信号　切入点：特征提取　出处：《自动化学报》2017年11期 　论文类型：期刊论文

【摘要】：正则化共空间模式(Regularized common spatial pattern,RCSP)解决了共空间模式(Common spatial pattern,CSP)对噪声敏感的问题,但它在小样本脑电数据集中的表现并不理想.针对上述问题,本文提出了Bagging RCSP(BRCSP)算法,通过Bagging方法重复选取训练数据来构造一个个包,并提取RCSP特征,再利用线性判别分析(Linear discriminant analysis,LDA)将特征向量映射到低维空间中,最后采用最近邻(Nearest neighborhood classifier,NNC)算法判定分类结果.线下实验证明,相比较聚合正则化共空间模式(RCSP with aggregation,RCSP-A),BRCSP的平均准确率提高了2.92%,且方差更小,鲁棒性更好.最后,在智能轮椅平台上,10位受试者利用BRCSP算法实现左右手运动想象脑电信号控制轮椅完成"8"字形路径的实验,证明了该算法在脑电信号特征提取中的有效性.
[Abstract]:Regularized common spatial pattern (RCSP) has solved the noise sensitivity problem of Common spatial pattern (CSP), but its performance in small sample EEG data sets is not satisfactory. In view of the above problems, a Bagging RCSP algorithm is proposed in this paper. The Bagging method is used to repeatedly select training data to construct packets and extract RCSP features. Then linear discriminant analysis (LDA) is used to map feature vectors to low dimensional space. In the end, the nearest neighbor neighborhood classifier is used to judge the classification results. The offline experiments show that the average accuracy of RCSP with aggregation is increased by 2.92%, and the variance is smaller and the robustness is better than that of RCSP with aggregation. On the platform of intelligent wheelchair, 10 subjects use BRCSP algorithm to realize the left and right hand motion imagination EEG control wheelchair to complete the "8" zigzag path experiment, which proves the effectiveness of the algorithm in EEG feature extraction.
【作者单位】：重庆邮电大学先进制造工程学院;重庆邮电大学自动化学院;
【基金】：重庆市科学技术委员会项目(cstc2015jcyjBX0066,cstc2017jcyjAX0033) 重庆市教委科学技术项目(KJ1600428)资助~~
【分类号】：R318;TN911.7

【相似文献】