一种改进的邻域多粒度粗糙集模型  

Improved neighborhood multi-granulation rough set model

作　　者：刘丹 徐立新 孟慧丽[2] 朱命冬 LIU Dan;XU Lixin;MENG Huili;ZHU Mingdong(Department of Computer Science ＆ Technology,Henan Institute of Technology,Xinxiang,Henan 453003,China;College of Computer and Information Engineering,Henan Normal University,Xinxiang,Henan 453007,China)

机构地区：[1]河南工学院计算机科学与技术系,河南新乡453003 [2]河南师范大学计算机与信息工程学院,河南新乡453007

出　　处：《计算机工程与应用》2018年第14期89-94,共6页Computer Engineering and Applications

基　　金：河南省高等学校重点科研项目(No.15A520062)

摘　　要：目前,邻域多粒度粗糙集模型广泛采用的距离函数闵可夫斯基距离存在着一定的局限性,通过引入兰氏距离作为距离函数,重构了邻域半径的选取方法,基于此提出一种改进的邻域多粒度粗糙集模型,并证明了相关的性质。采用UCI标准库数据集进行实验分析,对比两种模型的实验结果,验证了改进邻域多粒度粗糙集模型在近似逼近方面的优越性。At present, the neighborhood multi-granulation rough set widely adopting the Minkowski function is limited to some extend. In this paper, Lance distance is cited as the distance function, and reconstructs the neighborhood radius method. Then, an improved neighborhood rough multi-granulation rough set model is defined. In addition, some related properties are proven. The results of two models are compared through the experimental analysis on the UCI standard data sets, and also the superiority of the improved neighborhood multi-granulation rough set model is verified in terms of approximation approach.

关 键 词：粗糙集 邻域关系 多粒度 距离函数 兰氏距离 邻域半径 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]