一种基于模糊理论和条件熵的属性近似约简的方法  

An Algorithm of Approximate Reduction about Attribute Based on Fuzzy Theory and Conditional Information Entropy

作　　者：皋军[1] 王建东[1] 

机构地区：[1]南京航空航天大学计算机科学与工程系

出　　处：《计算机工程与应用》2004年第21期182-184,212,共4页Computer Engineering and Applications

摘　　要：给出了一种基于信息系统中连续型属性的模糊相似关系的定义以及相对应的关系矩阵,为了降低计算量对Warshall算法进行了改进。从信息论的角度提出了基于条件信息熵的属性新的近似相对约简集的概念和对应的约简算法,分析了算法的复杂度。实例和算法比较说明该算法是有效的。This paper produces a definition of fuzzy similarity relation and corresponding similarity matrix based on continuous-valued attribute in the information system.In order to reduce the amount of calculation,it improves Warshall algorithm.It also gives a new definition of relative approximate reduction and corresponding approximate algorithm from the angle of conditional information entropy,and meanwhile,analyzes the complexity of algorithm.At last,it proves the validity of above-mentioned algorithm through an example.

关 键 词：模糊集 粗糙集 模糊相似关系 WARSHALL算法 近似约简 

分 类 号：O159[理学—数学]