一种用于识别神经网络元件之间互连的新统计方法。

A new statistical method for identifying interconnections between neuronal network elements.

作者信息

Borisyuk G N, Borisyuk R M, Kirillov A B, Kovalenko E I, Kryukov V I

出版信息

Biol Cybern. 1985;52(5):301-6. doi: 10.1007/BF00355752.

Abstract

A new method is proposed to analyse dependencies in point processes, which takes into account specific character of neuronal activity. Simulation modelling of neuronal network revealed that the estimated weight of connection depends monotonically on the value of the model synaptic strength. In contrast to the crosscorrelation, the method allows for nonlinear interconnections and does not require point processes to be stationary and samples to be large. Examples are presented of the method's application to neurophysiological data analysis.

摘要

提出了一种分析点过程中相关性的新方法，该方法考虑了神经元活动的特定特征。神经网络的模拟建模表明，估计的连接权重单调依赖于模型突触强度的值。与互相关不同，该方法允许非线性互连，并且不要求点过程是平稳的，也不要求样本量大。文中给出了该方法应用于神经生理数据分析的示例。