摘要
图模型主要借助拓扑图的直观形式对多维概率分布进行统计推断和因果分析.利用图的可分解性,可将高维问题分解成低维问题,并借助图的直观结构对随机变量间复杂的独立性关系、时序关系或因果关系进行分析.本文介绍了图模型在统计建模中的应用.实例表明,图模型通过Markov性,可以把一些复杂的高维系统分解成若干简单的部分,对各个子系统进行分别处理,从而将问题简化.
Graph model utilizes the visual form of topological graphs to make statistical inferences and causal analysis for multiple probabilities. By using the decomposability of graph, high-dimensional problem can be decomposed into some low-dimensional problems, and then we can analyze the independence relation, time sequence relation and causal relation among random variables by means of visual structure of graph. This paper introduced modeling approaches and applications of graphical models. The result of the experiments showed that the high-dimensional complex systems can be decomposed into some simple components by the Markov properties of graphical model, and handle each subsystem respectively, thus simplify the problem.
出处
《周口师范学院学报》
CAS
2009年第5期112-116,共5页
Journal of Zhoukou Normal University
关键词
统计建模
图模型
MARKOV性
复杂问题
statistical modelling
graphical model
markov property
complex problem
