摘要
为了模拟动物大脑皮层结构连接与功能连接间的关系,为机器学习提供新的思路,本文用图论测度表示网络的结构,用信息论测度表示功能整合和功能分离间的相互作用,采用图选择的方法对随机图进行变异和选择,确定与特定的功能动力学模式相对应的网络结构,并研究在外界刺激信号作用下,系统连接与感觉层间的匹配关系。仿真结果表明:由图选择获得的网络结构,呈现若干密集的顶点区,区域间松散连接,具有功能分离与功能整合的特点。在外界信号刺激下,特定的结构模式可以使系统与感觉层信号的统计结构间达到最大程度的匹配。
To simulate the anatomical and functional connectivity of higher vertebrates cerebral cortex and afford a new way of machine learning, it is represented networks structure with graphs and described functional connectivity with information-theoretical measures. Then random graphs are selected and mutated using graph selection method to determine the relationship between structural features of networks (expressed as graphs) and the patterns of functional connectivity to which they give rise when implemented as dynamical systems. It is also measured how well the intrinsic correlations matches the statistical structure of the system sensory input. The simulation results show that the graphs decided by graph selection present dense groups of vertices linked by a relatively small number of reciprocal bridges. This pattern is consistent with the emergence of a functionally segregated and functionally integrated system. It also shows that special structure could fit the sensory statistical structure well as it responds to extrinsic stimulus.
出处
《系统仿真学报》
EI
CAS
CSCD
北大核心
2005年第2期510-512,共3页
Journal of System Simulation
基金
国家自然科学基金 (60375017)
关键词
结构连接
功能连接
图选择
复杂度
anatomical connectivity
functional connectivity
graph selection
complexity
