基于鼠脑海马位置细胞与Q学习面向目标导航  

Goal oriented navigation based on place cells of rat's brain hippocampus and Q-learning

作　　者：方略 何洪军[1] FANG Lue;HE Hongjun(The Twenty-First Research Institute of China Electronics Technology Group Corporation, Shanghai 200030, China)

机构地区：[1]中国电子科技集团公司第二十一研究所,上海200030

出　　处：《生物信息学》2019年第1期31-38,共8页Chinese Journal of Bioinformatics

基　　金：中国电子科技集团公司第二十一研究所资助项目(No.CB066)

摘　　要：生理学实验表明在鼠脑海马结构中存在的一种具有特异性放电特征的细胞,它在大鼠空间导航和环境认知中起着关键性作用,该特异性神经元被称之为位置细胞。本文将基于位置细胞、运动神经元来构建一种前馈神经网络模型,采用Q学习算法实现大鼠面向目标导航任务。实验结果表明该前馈神经网络模型能快速实现大鼠面向目标导航任务。Physiological experiments show that a cell with specific discharge characteristics in the hippocampus of rat s brain plays a key role in rat's spatial navigation and environmental cognition, and the specific neurons are called place cells. In this paper, a feed-forward neural network model based on place cells and motor neurons is constructed, and Q-learning algorithm is used to realize the goal-oriented navigation task of the rat. The experimental results show that the feed-forward neural network model can quickly achieve the goal-oriented navigation task of the rat.

关 键 词：位置细胞 Q学习算法 前馈神经网络模型 目标导航 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]