改进的启发式离散化算法  

Improved heuristic algorithm for discretization

作　　者：彭佳文[1] 覃纪武[2] 

机构地区：[1]广西工学院信息网络中心,广西柳州545006 [2]柳州职业技术学院,广西柳州545006

出　　处：《计算机工程与设计》2008年第15期4003-4005,共3页Computer Engineering and Design

基　　金：湖南省科技厅科技计划基金项目(05JT1013)

摘　　要：Nguyen S.H离散化算法中定义的初始断点集由于可能包含了部分对决策系统的分辨关系并无贡献的断点而影响到算法的效率。通过定义分界点来对该算法中定义的初始断点以对决策系统的分辨关系是否有贡献来进行区分,并仅取分界点集作为初始断点集,使得初始断点数目较大幅度地降低,提出了一种改进的启发式离散化算法。此算法较大程度地减小了算法空间复杂性和时间复杂性,对比实验结果表明了改进算法的正确性和有效性。It is possible that some dividing points included in the original breaking set defined by Nguyen S. H discretization algorithm have no contribution to the differentiation relation of the decision system, and this can affect the efficiency of the algorithm. By defining dividing point, each of the breaking points defined by Nguyen S.H discretization algorithm could be differentiated whether has contribution to the differentiation relation of the decision system. And by choosing the dividing point set as the original breaking set only, the number of original breaking point could be decrease to a biggish degree. An improved heuristic algorithm for discretization is introduced. The improved algorithm can reduce preferably the space complexity and time complexity of the discretization. The result of contrast experiment proves the correctness and validity of the improved algorithm.

关 键 词：粗糙集 离散化 决策表 分辨关系 分界点 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]