|本期目录/Table of Contents|

 基于支持向量特征筛选方法的想象动作识别(PDF)

《纳米技术与精密工程》[ISSN:1672-6030/CN:12-1351/O3]

期数:
2012年4期
页码:
348-354
栏目:
精密测量
出版日期:
2012-07-15

文章信息/Info

Title:
 Motor Imagery Recognition Based on Support Vector Feature Selection Method
作者:
 綦宏志12 明东2 万柏坤2 任超世1 刘志朋1 殷涛1
 (1. 中国医学科学院生物医学工程研究所,天津300192;
2. 天津大学精密仪器与光电子工程学院,天津300072)
Author(s):
 QI Hongzhi12 MING Dong2 WAN Baikun2 REN Chaoshi1 LIU Zhipeng1 YIN Tao1
 (1. Institute of Biomedical Engineering, Chinese Academy of Medical Sciences and Peking Union Medical College, Tianjin 300192, China;
2. School of Precision Instrument and Opto-Electronics Engineering, Tianjin University, Tianjin 300072, China)
关键词:
 支持向量特征筛选想象动作脑-机接口
Keywords:
 support vector feature selection imaginary motor brain-computer interface
分类号:
-
DOI:
-
文献标识码:
A
摘要:
 引入了支持向量特征筛选方法,以克服基于想象动作诱发脑电特征的脑-机接口识别中,由于特征维度较高而训练数据有限、不易获得理想识别效果的问题.支持向量特征筛选方法采用扰动支持向量机代价函数的方法测量特征的分类贡献度,进而建立特征序贯指数,以递归方法进行特征排序和优化筛选.对14例受试者的左右上肢想象动作诱发脑电信号进行分析,提取6类246维特征,采用支持向量递归筛选方法进行特征优选,利用支持向量机对优选特征进行识别,结果显示,支持向量递归筛选得到的优选特征可显著提高识别正确率.研究表明,支持向量特征筛选可以降低无效特征干扰,提高分类器效率,适用于特征维度较高的脑-机接口任务识别.
Abstract:
 This paper introduces a support vector feature selection method to improve the recognition of the motor imagery in brain-computer interface, in which it is usually hard to achieve a satisfactory result due to the massive feature dimension and the limited training data. Support vector feature selection measures the contribution of each feature to classification by disturbing the objective function of SVM. Then it constructs a feature ranking criteria and recursively ranks all features, and finally it selects the optimal feature group. Evoked potential induced by left versus right upper limb imaginary motor from 14 subjects is analyzed in this paper. Overall 246 features from 6 species are extracted and then optimized by support vector recursive feature selection. The classification result obtained by employing support vector machine shows that the optimized feature group improves accuracy significantly. This study indicates that the support vector feature selection method is capable of reducing the influence from redundant features and improving recognition efficiency, especially in the high feature dimension situation of brain-computer interface.

参考文献/References

备注/Memo

备注/Memo:
更新日期/Last Update: 2012-11-14