Abstract: Kullback-Leibler distance of information can be used to measure the difference degree between the random variables. This paper presents the concept of extended Kullback-Leibler distance. A new reduction algorithm based on extended Kullback-Leibler distance for knowledge is put forward, and the complexity of the algorithm is analyzed simply. The experimental results show that this algorithm can find out the relative minimal reduction for most decision tables.

Key words: Rough Set theory, Information theory, Kullback-Leibler distance, Attribute reduction

摘要： 信息的Kullback-Leibler距离能够很好地表示两个分布函数所包含信息的差异程度。文章提出了扩展Kullback-Leibler距离的概念，在此基础上提出了一种新的基于扩展Kullback-Leibler距离的属性约简算法。实验分析表明，在多数情况下该算法能够得到决策表的最小相对约简，同时还对算法复杂度作了简单分析。

关键词: Rough Set理论, 信息论, Kullback-Leibler距离, 属性约简

CLC Number: 

• TP311