基于邻域辨识矩阵的属性约简增量式算法  

Incremental updating algorithm for neighborhood-based attribute reduction based on discernibility matrix

作　　者：林俊伟[1] 叶东毅[1] 

机构地区：[1]福州大学计算机科学与技术系,福州350002

出　　处：《计算机应用》2009年第B06期119-121,共3页journal of Computer Applications

基　　金：国家自然科学基金资助项目(60602052);福建省自然科学基金资助项目(A0610014)

摘　　要：邻域粗糙集模型可以直接处理连续型数据。属性约简是邻域粗糙集模型的重要研究内容之一。目前已有的增量式粗糙集约简算法主要考虑经典的粗糙集模型的情况,不适用于邻域粗糙集模型。为此,提出一种基于邻域辨识矩阵的属性约简增量式算法,主要考虑连续型数据对象动态增加的情况下的属性约简问题。为了解决该问题,提出了一种邻域决策系统的辨识矩阵,通过辨识矩阵的动态更新实现动态求核,并在此基础上,利用原有的属性约简进行属性约简的更新。Neighborhood-based rough set model is able to handle continuous data directly. Attribute reduction is of primary importance in the context of this extended rough set model Many exiting incremental updating algorithms for attribute reduction aim at the case of classic rough set, but they are not suitable for neighborhood-based rough set. Therefore the author introduced an algorithm for neighborhood-based rough set. This algorithm based on the updating of neighborhood discernibility matrix could dynamically computed a new core. Base on this, attribute reduction could be effectively updated by utilizing the old attribute reduction.

关 键 词：邻域粗糙集 属性约简 核 增量式更新 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]