项目名称: 神经高斯随机场的建模、分析及应用
项目编号: No.61273309
项目类型: 面上项目
立项/批准年度: 2013
项目学科: 自动化技术、计算机技术
项目作者: 卢文联
作者单位: 复旦大学
项目金额: 60万元
中文摘要: 本项目的目标是基于计算神经模型,通过构建输入到输出神经放电点过程的激发率的期望、方差(变异系数)和相关系数的映射,建立新的神经高斯随机场模型,为研究计算神经科学建立一个新平台,利用数学和统计的工具分析计算神经科学中一些热点问题,并应用于分析生物数据。我们希望做到:(1)、利用相关系数的发展方程研究神经同步发放链的演化和时空分布;(2)、基于高斯场和信息理论分析神经网络中的信息传递;(3)、利用Young测度和放松优化理论研究如何在噪音环境中实现精确的神经控制 ;(4)、发展高斯随机场的学习算法,通过输入输出放电信号的相关系数研究STDP;(5)、基本建立基于神经高斯随机场的fMCI数据处理和分析框架。我们希望通过该模型的建立和研究,能提供一类模型和方法,为神经计算科学与人工神经网络理论之间隔阂找一个连接的桥梁。
中文关键词: 高斯神经场;协调性;神经-运动控制;基因影像;智能网络系统
英文摘要: The goal of the project is to establish a new neural Gaussian random field model by constructing input-output mapping of the expectations, variances (or equivalently coefficients of varation) and the coefficients of correlation of the spiking rates, based on the well-known computational neuronal models. By this way, we aim to construct a novel research platform, on which the well-developed mathematical and statistic tools can be utilised to study a number of hot topics in the computational neuroscience and treat biological data of neural activities. Our tasks include: (1). We are to investigate the spatial-temporal properties of neural synfire chain by studying the evolution and distribution of the correlation coefficients in the model; (2). We discuss the information transfer in the neural network via the Gaussian distributions of the neural dynamics; (3). We use the Young measure and relax optimization theory to realize precise neural control in a noisy circumstance; (4). We will develop learning algorithms of Gaussian random field, in particular a learning approach based on the input-ouput correlation to analyze STDP; (5). We are to provide a fundamental framework of the procssing and analyzing fMCI data based on neural Gaussian random field theory. We expect that the results of the project can provide a clas
英文关键词: Gaussian neural field;coordination;neuro-motion control;genetic-imaging;intelligent network systems