基于不可辨识矩阵的值约简算法  

An Algorithm for Value Reduction Based on Indiscernibility Matrix

作　　者：罗秋瑾[1] 成蓉华[1] 纳静[1] 

机构地区：[1]云南财经大学统计与数学学院,云南昆明650221

出　　处：《云南民族大学学报（自然科学版）》2011年第6期508-510,共3页Journal of Yunnan Minzu University:Natural Sciences Edition

摘　　要：属性值约简是粗糙集理论的重要研究课题之一,很多学者对它进行研究并提出了不同的算法,但由于值约简是NP-hard问题,目前还没有高效的方法.根据可辨识矩阵的定义,提出了不可辨识矩阵,将其运用到属性值约简的问题中.实验结果验证了此算法的可行性和有效性,能节省循环比较时间,提高计算速度.The value reduction is an important research topic in the rough set theory, and many researchers have studied it and put forward different algorithms of value reduction. Because the value reduction algorithm is a NP - hard problem, there is no effective algorithm at present. Based on the definition of identifiable matrix, this paper puts forward an indiscernibility matrix and applies it to the problem of value reduction. It is proved in the experi- ment that this algorithm saves time on circular comparing and increases computing speed.

关 键 词：粗糙集 不可辨识矩阵 值约简 规则 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]