关联粒的发育自适应挖掘  

Self-adaptive Mining for Well-developed Associated Granules

作　　者：谷峰[1] Gu Feng(Zhejiang Technical Institute of Economics,Hangzhou 310018,China))

机构地区：[1]浙江经济职业技术学院,杭州310018

出　　处：《科技通报》2017年第5期110-115,共6页Bulletin of Science and Technology

基　　金：浙江省教育科学规划课题(2015SCG300);2016年浙江省教育技术规划课题(JB116);杭州市哲学社会科学规划课题基地项目(2016JD50);教育部职业院校文化素质教育指导委员会"十二五"规划教育科研课题(WH125YB13);浙江经济职业技术学院教改课题(xjg2015002);浙江经济职业技术学院学生科技竞赛项目(201618)

摘　　要：从基于属性值域划分的不可区分关系和相容关系,建立全相容关系、广义相容关系和关联关系。由此得到关联粒,即广义相容粒在约束条件下的滤存。关联粒层包含度定义了关联粒的发育水平。属性值域的自由划分使粒度空间可以自由转换,帮助寻找最佳关联粒层。文中给出了关联粒和划分的同构约简。通过给出同域方差优化均值等值变换和一个基于Kappa一致性系数的鲁棒的相容函数,给出了具体的关联粒发育自适应挖掘算法。实例计算表明算法有效的挖掘了最佳关联粒层。From the domain division based indiscernibility and compatibility,fully compatibility,generalized compatibility,and associability were defined.So the associated granule is defined,which is aconstraint of a generalized compatible granule.The associated granule layer inclusion describesdevelopment of an associated granule layer.Arbitrariness of domain division makes the free transformingof associated granular spaces,and so easy to find the best associated granular layer.Isomorphicreduction of domain divisions and associated granules was presented.Based on an attributehomogenizing transformation and a kappa coefficient based robust compatible function,a self-adaptivewell-developed associated granule mining algorithm was designed.Experiments show that the algorithmmined well-developed associated granules effectively.

关 键 词：属性值域划分 广义相容 关联粒 粒层包含度 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]