单电极中潜伏期反应的听觉注意特征提取与识别
发布时间：2018-08-28 17:50
【摘要】：通过提取单电极中潜伏期反应(MLR)的特征差异,研究并实现了正常个体听觉注意与非注意2种状态的识别.首先,对MLR信号进行小波滤波、阈值去伪迹、相干平均等预处理;然后,分析了MLR在2种状态下的成分波差异,并将Na,Pa,Nb波的幅值与能量、面积、C0复杂度、AR模型系数等传统特征组合成为新的特征向量;最后,采用支持向量机(SVM)和人工神经网络(ANN)在传统特征向量和新特征向量下进行目标识别.8位被试的实验结果显示,在2种不同状态下,被试的Na,Pa,Nb波幅值具有显著性差异(p0.05),而潜伏期并无差异.ANN作为分类器时,新特征向量的平均识别正确率可达85.7%.由此可见,利用单电极中潜伏期反应区分听觉注意与非注意状态是有效的.
[Abstract]:By extracting the characteristic difference of latency response (MLR) in single electrode, the recognition of two states of auditory attention and non-attention in normal individuals was studied and realized. Firstly, the MLR signal is preprocessed with wavelet filtering, threshold de-artifact, coherent average, etc. Then, the difference of component wave in two states of MLR is analyzed, and the amplitude and energy of Na,Pa,Nb wave are analyzed. The traditional features such as area C0 complexity and AR model coefficients are combined into new feature vectors. Finally, Support vector machine (SVM) and artificial neural network (ANN) are used to recognize the target under traditional and new feature vectors. The experimental results show that in two different states, The amplitude of Na,Pa,Nb was significantly different (p0. 05), but there was no difference in latency. Ann was used as classifier, the average recognition accuracy of the new feature vector could reach 85.7%. Therefore, it is effective to distinguish auditory attention from non-attention by single-electrode mid-latency response.
【作者单位】： 东南大学信息科学与工程学院;广州大学机械与电气工程学院;南京工程学院通信工程学院;
【基金】：国家自然科学基金资助项目(61375028,61673108) 江苏省“六大人才高峰”资助项目(2016-DZXX-023) 江苏省博士后科研资助计划资助项目(1601011B) 江苏省“青蓝工程”资助项目 广州大学广东省灯光与声视频工程技术研究中心开放基金资助项目(KF201601,KF201602)
【分类号】：R318;TP18
，
本文编号：2210122
[Abstract]:By extracting the characteristic difference of latency response (MLR) in single electrode, the recognition of two states of auditory attention and non-attention in normal individuals was studied and realized. Firstly, the MLR signal is preprocessed with wavelet filtering, threshold de-artifact, coherent average, etc. Then, the difference of component wave in two states of MLR is analyzed, and the amplitude and energy of Na,Pa,Nb wave are analyzed. The traditional features such as area C0 complexity and AR model coefficients are combined into new feature vectors. Finally, Support vector machine (SVM) and artificial neural network (ANN) are used to recognize the target under traditional and new feature vectors. The experimental results show that in two different states, The amplitude of Na,Pa,Nb was significantly different (p0. 05), but there was no difference in latency. Ann was used as classifier, the average recognition accuracy of the new feature vector could reach 85.7%. Therefore, it is effective to distinguish auditory attention from non-attention by single-electrode mid-latency response.
【作者单位】： 东南大学信息科学与工程学院;广州大学机械与电气工程学院;南京工程学院通信工程学院;
【基金】：国家自然科学基金资助项目(61375028,61673108) 江苏省“六大人才高峰”资助项目(2016-DZXX-023) 江苏省博士后科研资助计划资助项目(1601011B) 江苏省“青蓝工程”资助项目 广州大学广东省灯光与声视频工程技术研究中心开放基金资助项目(KF201601,KF201602)
【分类号】：R318;TP18
，
本文编号：2210122
本文链接：https://www.wllwen.com/kejilunwen/zidonghuakongzhilunwen/2210122.html
