多标记的不完备多粒度决策系统的最优粒度选择  

Optimal Granularity Selections in Incomplete Multi-granular Decision Systems with Multi-decision Labels

作　　者：史进玲[1] 张江维[1] 程菊明[2] SHI Jinling;ZHANG Jiangwei;CHENG Juming(School of International Education,Xuchang University,Xuchang 461000,China;School of Information Engineering,Xuchang University,Xuchang 461000,China)

机构地区：[1]许昌学院国际教育学院,河南许昌461000 [2]许昌学院信息工程学院,河南许昌461000

出　　处：《许昌学院学报》2021年第2期103-108,共6页Journal of Xuchang University

基　　金：河南省科技攻关重点项目(192102210055);河南省高等学校重点科研项目(17B520036);2018年许昌市校地合作专项。

摘　　要：粒计算是针对复杂问题求解、海量数据挖掘的一个重要方法.在许多实际应用中,人们往往通过选择合适的粒度来解决问题.针对不完备信息系统的粒度选择问题,首先,引入了多标记的不完备多粒度决策系统,定义了多粒度视角下隶属度概念;其次,介绍了决策系统协调性判断标准,讨论了决策系统的全局最优粒度选择;最后,为弥补全局粒度选择不能使每个对象都达到最优粒度的局限性,给出了对象的局部最优粒度选择方法.Granular computing is an important method for solving complex problems and massive data mining.In many practical applications,people often solve the problem by choosing the appropriate granularity.Aiming at the problem of granularity selection in incomplete information systems,the multi-label incomplete multi-granular decision system is introduced firstly,and the concept of membership degree from multi-granularity levels is defined.Then,the consistent evaluation criterion of decision system is introduced,and the global optimal granularity selection methods of decision systems are discussed.Finally,in order to overcome the limitations that the global optimal granularity selection cannot make every object reach the optimal granularity,the local optimal granularity selections of objects are explored.

关 键 词：粒计算 多粒度 不完备信息系统 多标记决策 最优粒度 

分 类 号：TP182[自动化与计算机技术—控制理论与控制工程]