基于粗集的规则提取LBR和LEM3  

METHODS OF EXCAVATING RULES BASED ON ROUGH SET: LBR AND LEM3

作　　者：胡丹[1] 莫智文[1] 

机构地区：[1]四川师范大学数学研究所,成都610066

出　　处：《模式识别与人工智能》2002年第2期129-133,共5页Pattern Recognition and Artificial Intelligence

基　　金：国家自然科学基金(No.69803007)

摘　　要：本文基于粗集理论,提出了一种新的规则提取法LBR(Learning By Rough Sets),并对LBR与另一种已有的规则提取法LEM1,即全局覆盖算法(global covering algorithm)进行了比较和讨论.基于比较的结果,得出了将LEM1改进后的LEM3.LBR不但可用于普通的决策表规则提取,更多地可应用于基于模糊划分的规则提取.LBR的提出,极大地简化和丰富了规则提取算法,在已知数据中可获取更为丰富的信息量.而LEM3的使用,则是在将"依赖"(depend on)这一概念推广的基础上,更灵活地使用"覆盖"(covering),扩大了获取规则的范围.LBR和LEM3因其各自不同的优点,在数据挖掘和智能领域均具有广泛的应用前景.By the help of rough set theory, a new way of machine learing - LBR( Learning by Rough Set theory) is put forward in this paper and then, it is compared with LEM1 the algorithm proposed before. From the results of comparison, a new method of excavating rules from examples named LEM3, which is more flexible than LEM1 is found. The use of LBR simplifies the algorithms of machine learning and helps us to get more information from examples, while LEM3 expands the scope of rules gained basing on the extension of the concept 'depend on' and the flexible use of 'covering'. Because of their respective advantages, LBR and LEM3 have extensive applied prospect in the field of excavating information from examples.

关 键 词：粗集 规则提取 LBR LEM3 决策表 模糊划分 机器学习 人工智能 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]