纵向环正则规则长入与后缩的MMDR算法  

Implementing MMDR Using Deep-first Node-regularity Growing and Shrinking Algorithm

作　　者：林晓旻[1] 王治宝[1] 

机构地区：[1]南开大学信息技术科学学院,天津300071

出　　处：《计算机工程与应用》2005年第11期214-217,共4页Computer Engineering and Applications

摘　　要：利用复杂适应系统为证券系统建模,需要对大量的个体建立模型,如何用简单的方法生成大量模型,是一个值得研究的问题。该文对IF-THEN正则规则(regularity)引入了“正则规则集———符号概率推导链”的概念,证明了环正则规则长入与后缩的构造性定理,给出了环正则规则生枝、剪芽、长入与后缩的操作规则,完成了纵向环正则规则长入与后缩的MMDR(aMachineMethodforDiscoveringRegularities)算法。同时阐述了如何将该算法应用于证券交易访真复杂适应系统的“个体”建模中。Very large number of individual models need to be built to simulate stock market using CAS(Complex Adaptive System).It is a valuable issue to produce multitudinous models with a single method.This paper imports the concept of'regularity set-chains of semantic probabilistic inference'into IF-THEN regularity,and proves the constructive theorem of node-regularity growing and shrinking.Operation rules of branching,pruning,growing and shrinking for node-regularity are given in the paper.Finally the paper completes a deep -first node -regularity growing and shrinking algorithm to implement MMDR.At the end it introduces how to apply this algorithm to build individual models in stock market simulation.

关 键 词：复杂适应系统 证券交易市场仿真 MMDR算法 符号概率推导链 正则规则 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]