基于信鸽视顶盖神经集群响应的目标类别解码  

Decoding Object Class Based on Responses of a Population of Pigeons' Optic Tectum

作　　者：师黎[1] 胡汉顶 王松伟[2] 

机构地区：[1]清华大学自动化系,北京100084 [2]郑州大学电气工程学院,郑州450001

出　　处：《科学技术与工程》2017年第31期35-41,共7页Science Technology and Engineering

基　　金：国家自然科学基金(U1304602);河南省重点科技攻关计划(122102210102);河南省高校科技创新人才支持计划(124200510016);河南省高校科技创新团队支持计划(2012IRTSTHN001)资助

摘　　要：通过提取信鸽视顶盖(the optic tectum,OT)的中间层神经元集群高频峰电位(spike)响应信号,利用小波和信息论的方法提取并筛选不同时间尺度上的响应特征,构建集群响应的特征向量,从而实现对目标类别的解码。首先,根据感受野特征筛选有效响应通道;然后通过特定目标对信鸽进行刺激实验,采用微电极阵列记录多通道的spike发放序列;其次设定时间窗口(bin)长度,采用小波和信息论的方法提取并筛选响应特征;最后利用朴素贝叶斯分类算法实现目标类别解码。结果表明,一方面所提取的响应特征能够实现对目标类别的解码和识别;另一方面,通过对bin取1 ms的响应信号所提取特征进行统计分析,发现提取的包含目标类型信息高的小波系数大都属于较大分解层次上的系数,表明鸽子OT区可能利用这些尺度上的时间模式对目标的信息进行了编码,这为研究OT中间神经元的时间响应结构奠定了基础。By extracting the spike trains from the middle layer neural population of the optic tectum( OT) of the pigeon,the response characteristics of the effective response interval were extracted by wavelet and information theory,and the eigenvector of the population responses was constructed. Then the classification of identification of the object was realized. First,the effective response channels were selected according to the characteristics of the receptive field. Then,the stimulus experiments were carried out on the pigeon by the specific object,and the multi-channel spike trains were recorded simultaneously using a multielectrode array. Then,the effective response interval was selected,the value of the time window was set. The wavelet transform and information theory were used to filter and extract the response feature structure,and the cluster response feature vector was constructed. Finally,decoding of the object class was realized by using the naive Bayesian classification algorithm. The results showed that the response feature extracted by this method could realize object class decoding. Based on the statistical analysis of the extracted features with high information of object class,it is found that the wavelet coefficients of high wavelet decomposition level had a large proportion,which indicated that some features of the object could be coded by time patterns on these time scales by the OT neurons of pigeons. The foundation was laid for the research of time structure of the responses of OT middle layer neurons.

关 键 词：视顶盖 神经元集群 小波 信息论 特征矩阵 解码 

分 类 号：Q424[生物学—神经生物学]