摘要
为提高癫痫脑电(EEG)信号的正确识别率,设计了一种基于非线性特征提取的EEG信号支持向量分类器.分类器首先将EEG信号通过四层小波包变换分解到不同频段,然后计算各频段小波系数的近似熵(ApEn)值,作为特征向量,最后使用支持向量机(SVM)进行分类.实验结果显示该分类器能有效提高正确识别率.
A nonlinear feature extraction based SVMclassifier for EEG signals was proposed to improve the correct classification rates of epileptic EEG. Firstly, EEG signals were decomposed into various frequency bands through fourth-level wavelet packet decomposition. Secondly, approximate entropy（ApEn） values of the wavelet coefficients were used as feature vectors. Lastly, the SVM was used in classification.Experimental results showed that the correct classification rates of the classifier proposed was improved.
出处
《汕头大学学报（自然科学版）》
2009年第1期69-74,共6页
Journal of Shantou University：Natural Science Edition
