两类运动想象中无效采样对脑电特征提取影响的探讨
首发时间:2018-05-15
摘要:脑电样本通常含有无效采样,对脑电识别产生各种影响。本文以国际脑机接口竞赛中多个关于两类运动想象的脑电数据集为素材,利用小波变换、滑动平均等方法提取多种脑电信号特征,从样本的无效采样点比例变化、不同的脑电特征提取方法、不同的样本容量等多层面分析无效采样对脑电识别效果的影响。研究表明无效采样形成特征缺失,对各种特征提取方法产生不同的影响,包括脑电识别效果稳定性的降低,以及识别模型的完整性受影响等,然而样本容量的增加又可提升识别效果的稳定性。因此脑电特征提取方法的选取要考虑如无效采样含量、样本容量等特征提取方法以外的因素。
For information in English, please click here
Discussions on the Effect of Void Sampling to EEG Features Extracting in Two Types of Motor Imagery
Abstract:EEG samples usually contain void sampling data then make various effects on EEG identification. In the paper, the two-type motor imagery EEG data sets of the international BCI competitions was used as EEG samples, and the method of wavelet transform, moving average etc was used to extract various EEG features. From various aspects, including the proportion change of void sampling data, different methods of EEG feature extraction, and different sample size etc, the paper analyzed the effects of void sampling data on EEG identifying results. The study showed that void sampling data can cause feature missing, and then make different effects on different feature extraction methods, including decreased stability of EEG identification result, and the effect on the integrity of the identification model etc, however, the increase of sample size can increase the stability of the identification result. Therefore, to select the methods of EEG feature extraction, the factors of void sampling data content and sample size etc should be considered besides feature extraction methods.
Keywords: BEI (Biotic Electric Interface) EEG (electroencephalogram) wavelet transform sampling conformity rate
基金:
引用
No.****
动态公开评议
共计0人参与
勘误表
两类运动想象中无效采样对脑电特征提取影响的探讨
评论
全部评论0/1000