Classification of stationary neuronal activity according to its information rate
Classification of stationary neuronal activity according to its information rate
复制标题
DOI：
10.1080/09548980600594165
复制
发表时间：
2006-06-01
影响因子：
7.8
通讯作者：
Lansky, Petr
中科院分区：
文献类型：
--
作者：
Kostal, Lubomir;Lansky, Petr
我们提出了一个衡量的信息率的一个单一的固定的神经元活动的空信息的状态。该措施是基于两个峰间期分布之间的Kullback-Leibler距离。所选择的活动与泊松模型具有相同的平均放电频率进行比较。我们表明，该方法与特定信息的概念有关，并且该方法允许我们判断相对编码效率。根据神经元活动模型的信息率，将其分为两类：更新过程模型和一阶马尔可夫链模型。已经证明，信息可以被传输，既不改变尖峰速率也不改变变异系数，并且序列相关性的增加不一定增加信息增益。我们采用简单，但功能强大，Vasicek的估计微分熵来说明应用程序的实验数据来自大鼠的嗅觉感觉神经元。
We propose a measure of the information rate of a single stationary neuronal activity with respect to the state of null information. The measure is based on the Kullback-Leibler distance between two interspike interval distributions. The selected activity is compared with the Poisson model with the same mean firing frequency. We show that the approach is related to the notion of specific information and that the method allows us to judge the relative encoding efficiency. Two classes of neuronal activity models are classified according to their information rate: the renewal process models and the first-order Markov chain models. It has been proven that information can be transmitted changing neither the spike rate nor the coefficient of variation and that the increase in serial correlation does not necessarily increase the information gain. We employ the simple, but powerful, Vasicek's estimator of differential entropy to illustrate an application on the experimental data coming from olfactory sensory neurons of rats.