[1]陈佳卉,王友国,翟其清.基于 K 近邻的运动想象分类中的噪声效益[J].计算机技术与发展,2022,32(01):79-84.[doi:10. 3969 / j. issn. 1673-629X. 2022. 01. 014]
　CHEN Jia-hui,WANG You-guo,ZHAI Qi-qing.Noise Benefits in Motor Imagery Classification UsingK-nearest neighbor[J].,2022,32(01):79-84.[doi:10. 3969 / j. issn. 1673-629X. 2022. 01. 014]
点击复制

基于 K 近邻的运动想象分类中的噪声效益()
分享到：

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:
32
期数:
2022年01期
页码:
79-84
栏目:
大数据分析与挖掘
出版日期:
2022-01-10

文章信息/Info

Title:
Noise Benefits in Motor Imagery Classification UsingK-nearest neighbor
文章编号:
1673-629X(2022)01-0079-06
作者:
陈佳卉; 王友国; 翟其清
南京邮电大学 理学院,江苏 南京 210023
Author(s):
CHEN Jia-hui; WANG You-guo; ZHAI Qi-qing
School of Science,Nanjing University of Posts and Telecommunications,Nanjing 210023,China
关键词:
随机共振; 高斯噪声; K 近邻; 运动想象; 脑电信号
Keywords:
stochastic resonance; Gaussian noise; K-nearest neighbor; motor imagery; EEG signal
分类号:
TP18
DOI:
10. 3969 / j. issn. 1673-629X. 2022. 01. 014
摘要:
关于脑电信号中的噪声处理问题一直是脑-机接口( BCI)领域的重点研究方向,通常认为噪声是有害的,所以针对脑电信号中的噪声处理往往以降噪或消噪为主。 但是根据随机共振( SR) 的思想,在非线性系统中噪声往往能增强信号处理,而脑电信号恰好具有非线性的特征,因此提出运用高斯噪声提高运动想象脑电信号的识别率。 通过在脑电信号中加入独立的高斯噪声,将原始训练集与添加噪声的训练集串联起来增加训练样本量,考虑训练样本量增加与否和噪声加入的阶段( 训练或 / 和测试) ;通过共空间模式( CSP)和小波包变换( WPT) 提取分类特征,并用 K 近邻( KNN) 算法进行分类。实验结果表明,只要加入适当强度的噪声,均可提高系统的分类准确率,出现随机共振现象;增加训练样本量的同时在训练集和测试集中加入适当强度相同的噪声,系统最大平均分类准确率相比不加噪声时增加 9. 28 个百分点;K 近邻算法的最大平均分类准确率相比决策树( DT)和支持向量机( SVM) 而言整体更高,体现出 K 近邻算法的优越性和可靠性。
Abstract:
The problem of noise processing in EEG signal has always been a key research direction in the field of brain-computer interface( BCI) ,and it is usually considered that noise is harmful, so the noise processing in EEG signal often focus on noise reduction orcancellation. However, according to the principle? ?of stochastic resonance ( SR ) , noise can enhance signal processing in nonlinearsystems,and EEG signals have nonlinear characteristics. Therefore, Gaussian noise is used to improve the recognition rate of motorimaginary EEG signals. Adding independent Gaussian noise to EEG signals, the original training set connects with the noise - addedtraining set to increase the training sample size. Considering whether training sample size increases or not and the stage of noise addition( training or / and testing) . The classification features are extracted by common spatial pattern ( CSP) and wavelet packet transform( WPT) ,and classified by K-nearest neighbor ( KNN) . The experimental results show that the accuracy of classification system can beimproved and appear stochastic resonance as long as add the appropriate noise intensity. When increasing the training sample size andadding the appropriate noise of the same intensity to the training set and test set,the maximum average classification accuracy of system is9. 28 percentage points higher than that without noise. The maximum average classification accuracy of K-nearest neighbor is higher thanthat of decision tree ( DT) and support vector machine? ? ? ? ( SVM) ,which reflects the superiority and reliability of K-nearest neighbor.

相似文献/References:

[1]邱勤伟 焦贤发 朱良燕.对称二值噪声下线性系统的随机共振[J].计算机技术与发展,2010,(08):239.
　QIU Qin-wei,JIAO Xian-fa,ZHU Liang-yan.Stochastic Resonance in Linear System with Dichotomous Noise[J].,2010,(01):239.
[2]王友国 刘沁雨.多阈值系统中高斯混合噪声改善信息的传输[J].计算机技术与发展,2011,(04):120.
　WANG You-guo,LIU Qin-yu.Gaussian Mixture Noise to Improve Information Transmission in Multi-threshold System[J].,2011,(01):120.
[3]翟其清,王友国,郑克. 噪声改善码元传输[J].计算机技术与发展,2014,24(11):152.
　ZHAI Qi-qing,WANG You-guo,ZHENG Ke. Noise Improving Code Elements Transmission[J].,2014,24(01):152.
[4]沈德海,张龙昌,鄂旭,等. 基于多子窗口的混合噪声滤波算法[J].计算机技术与发展,2015,25(06):69.
　SHEN De-hai,ZHANG Long-chang,E Xu,et al. A Mixture Noise Filter Algorithm Based on Multiple Sub-windows[J].,2015,25(01):69.
[5]李晓飞,邱晓晖. 基于小波变换的改进软阈值图像去噪算法[J].计算机技术与发展,2016,26(05):76.
　LI Xiao-fei,QIU Xiao-hui. An Improved Soft-threshold Image Denoising Algorithm Based on Wavelet Transform[J].,2016,26(01):76.
[6]黄玲俐. 一种改进权重的非局部均值图像去噪方法[J].计算机技术与发展,2016,26(06):16.
　HUANG Ling-li. A Non-local Means Denoising Algorithm with Improved Weighted Function[J].,2016,26(01):16.
[7]王友国[][],潘慧[],刘健[]. 加性乘性噪声改善多元信号检测[J].计算机技术与发展,2016,26(10):160.
　WANG You-guo[][],PAN Hui[],LIU Jian[]. Improvement of Multiple Signal Detection by Additive and Multiplicative Noise[J].,2016,26(01):160.
[8]徐龙飞,郁进明.不同优化器在高斯噪声下对 LR 性能影响的研究[J].计算机技术与发展,2020,30(03):7.[doi:10. 3969 / j. issn. 1673-629X. 2020. 03. 002]
　XU Long-fei,YU Jin-ming.Study on Influence of Different Optimizers on Performance of LR under Gaussian Noise[J].,2020,30(01):7.[doi:10. 3969 / j. issn. 1673-629X. 2020. 03. 002]
[9]黄 慧,董林鹭,何建华,等.强噪声下改进 Canny 算法的边缘检测[J].计算机技术与发展,2021,31(01):83.[doi:10. 3969 / j. issn. 1673-629X. 2021. 01. 015]
　HUANG Hui,DONG Lin-lu,HE Jian-hua,et al.Edge Detection of an Improved Canny Algorithm under Strong Noise[J].,2021,31(01):83.[doi:10. 3969 / j. issn. 1673-629X. 2021. 01. 015]

更新日期/Last Update: 2022-01-10