基于粗糙集理论的特征提取方法  

A Algorithm for Attribution Reduction Based on Rough Set Theory

作　　者：杜茜[1] 杜晓[2] 

机构地区：[1]临沂师范学院信息学院,山东临沂276005 [2]江苏科技大学电子信息学院,江苏镇江212003

出　　处：《科技信息》2007年第27期61-61,63,共2页Science & Technology Information

摘　　要：约简是粗集理论的重要概念,由定义计算约简是一个典型的NP问题且由于约简的不唯一,往往最后获得的属性集并非是最好的属性。本文针对Rought set理论的属性约简进行了研究。研究了通过可辨识矩阵求得属性约简集,利用Rought set与灰色理论相结合,提出一种属性约简的启发式算法。实验证明可以获得约简集合中最有利的属性。Reduction is an important concept in rough set theory, while computing reduction according to the definitions directly is a typical NP problem. This paper discusses the approaches for attribute reduction based on rough set theory. Following it studies the approaches to achieve attribute reduction set by applying recognized matrix, The text has researehed how to get attribute reductions through Discernibility matrix, combine Rough sets and gray theory, and puts forward a new heuristics algorithm for attribute reduction. The effectiveness of the result obtained is demonstrated by an example.

关 键 词：Rought SET理论 属性约简 分辨矩阵 灰色关联 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]