脑-机接口中基于ERS/ERD的自适应空间滤波算法

吕俊; 谢胜利; 章晋龙

doi:10.3724/SP.J.1146.2007.01462

脑-机接口中基于ERS/ERD的自适应空间滤波算法

doi: 10.3724/SP.J.1146.2007.01462

基金项目:

国家自然科学基金重点项目(U0635001)和国家自然科学基金项目(60505005，60774094)资助课题

计量
- 文章访问数: 3846
- HTML全文浏览量: 80
- PDF下载量: 1627
- 被引次数: 0
出版历程
- 收稿日期: 2007-09-13
- 修回日期: 2008-04-08
- 刊出日期: 2009-02-19

Adaptive Spatial Filter Based on ERD/ERS for Brain-Computer Interfaces

摘要

摘要: 在与运动相关的脑-机接口(Brain-Computer Interface, BCI)研究中，如果样本规模小，共同空间模式(Common Spatial Patterns, CSP)滤波算法对离群点(可能为噪声)敏感，鲁棒性不好。为此该文提出自适应空间滤波(Adaptive Spatial Filter, ASF)算法，抽取滤波后脑电信号的方差作为特征，并寻找最优滤波器使两类特征中心的比值最大。与CSP不同，ASF是迭代算法，具有软判决机制，能够依据历代更新后的滤波器，自适应地降低离群点对各类特征中心计算带来的影响。采用BCI competition 2003和2005中两套数据集进行实验，结果表明：尤其是在训练样本少的情况下，相对于CSP，ASF所提取的特征分类效果更好。
- 脑-机接口(BCI);特征提取;共同空间模式(CSP)滤波法
Abstract: For motor related Brain-Computer Interface (BCI), if the sample size is small, Common Spatial Patterns (CSP) algorithm is sensitive to outlier data and lacks of robustness. In this paper, an Adaptive Spatial Filter (ASF) algorithm is proposed to take filtered samples variances as the features and seek the spatial filter to maximize the ratio of two classes means. Unlike CSP, ASF is an iterative algorithm and have soft determination. ASF can adaptively decrease outliers effects according to the updated filters. Using two datasets from BCI competition 2003 and 2005, the experimental results show that ASF outperforms CSP，especially when training samples are few.

HTML全文

参考文献(1)

Lemm S, Schafer C, and Curio G. BCI competition 2003-dataset III: Probabilistic modeling of sensorimotor rhythms forclassification of imaginary hand movements [J].IEEE Trans.on Biomedical Engineering.2004, 51(6):1077-1080[2]Li Y Q and Guan C T. A semi-supervised SVM learningalgorithm for joint feature extraction and classification inbrain computer interfaces [C]. The 28th Annual InternationalConference of the IEEE Engineering in Medicine and BiologySociety, New York, USA, Aug.30-Sep.3, 2006: 2570-2573.[3]Lemm S, Blankertz B, and Curio G, et al.. Spatio-spectralfilters for improving the classification of single trial EEG [J].IEEE Trans. on Biomedical Engineering.2005, 52(9):1541-1548[4]McFarland D J, Anderson C W, and Mller K R, et al.. BCImeeting 2005-workshop on BCI signal processing: Featureextraction and translation [J].IEEE Trans. on Neural andRehabilitation Systems Engineering.2006, 14(2):135-138[5]Hammon P S and deSa V R. Preprocessing and metaclassification for brain- computer interfaces [J].IEEE Trans.on Biomedical Engineering.2007, 54(3):518-525[6]Wang Y J, Zhang Z G, and Li Y, et al.. BCI competition2003-data set IV: An algorithm based in CSSD and FDA forclassifying single-trial EEG [J]. IEEE Trans. on BiomedicalEngineering, 2004, 51(6): 1081-1086.[7]Liao X, Yao D Z, and Wu D, et al.. Combining spatial filtersfor the classification of single-trial EEG in a finger movementtask [J].IEEE Trans. on Biomedical Engineering.2007, 54(5):821-831[8]Friedman J, Hastie T, and Tibshirani R. Additive logisticregression: A statistical view of boosting [J]. The Annals ofStatistics, 2000, 28(2): 337-407.[9]Blankertz B, Mller K R, and Curio G, et al.. The BCIcompetition 2003: Progress and perspectives in detection anddiscrimination of EEG single trials [J].IEEE Trans. onBiomedical Engineering.2004, 51(6):1044-1051[10]Wei Q G, Gao X G, and Gao S K. Feature extraction andsubset selection for classifying single-trial ECoG duringmotor imagery [C]. The 28th Annual InternationalConference of the IEEE Engineering in Medicine and BiologySociety, New York, USA, Aug.30-Sep.3, 2006: 1589-1592.[11]Atkeson C G, Moore A W, and Schaal S. Locally weightedlearning [J].Artificial Intelligence Review.1997, 11(15):11-73[12]Yang J and Yang J Y. Why can LDA be performed in PCAtransformed space [J].Pattern Recognition.2003, 36(12):563-566[13]Sun Y J. Iterative RELIEF for feature weighting: algorithms,theories, and applications [J].IEEE Trans. on PatternAnalysis and Machine Intelligence.2007, 29(6):1035-1051

施引文献

资源附件(0)

访问统计