对象集变化时相容块的近似集增量更新方法  

An incremental method for updating approximations of consistent blocks while the universe evolves over time

作　　者：刘斌 王磊[1,2] 王冲 蔡香香 LIU Bin;WANG Lei;WANG Chong;CAI Xiangxiang(School of Information Engineering,Nanchang Institute of Technology,Nanchang 330099,Jiangxi,China;Jiangxi Provincial Key Laboratory of Water Information Cooperative Sensing and Intelligent Processing,Nanchang 330099,Jiangxi,China)

机构地区：[1]南昌工程学院信息工程学院,江西南昌330099 [2]江西省水信息协同感知与智能处理重点实验室,江西南昌330099

出　　处：《山东大学学报（工学版）》2023年第2期109-117,共9页Journal of Shandong University（Engineering Science）

基　　金：江西省教育厅科技项目(GJJ170995);国家自然科学基金项目(61562061)。

摘　　要：为解决相容块粗糙集模型中概念近似集静态算法效率低的问题,对象集发生变化时,从新的角度提出一种增量式近似集更新方法。引入相容块粗糙集模型的基本概念,在对象集发生变化的条件下更新论域子集,结合增加(或删除)的对象集与相容块之间的联系,更新每一个相容块,并提出更新相容块在论域上覆盖和划分的方法,设计一种近似集增量式更新算法。6组UCI数据集试验结果表明,在增加对象时,本研究算法比非增量式更新算法时间缩短了1/3,在删除对象时时间缩短了1/2。在同种数据集下,本研究算法的时间消耗均小于其他4种更新算法,验证了算法的高效性和可行性。To tackle the low efficiency of static algorithms of conceptual approximations in the consistent block rough set model,an incremental approximation updating method was proposed from a new perspective in case the object set varies.The fundamental con-cept of consistent block rough set model was introduced;the universe subset was updated when the object set varies;each consistent block was updated with the association between the adding(or deleting)object set and the consistent block,then we proposed a method of updating the coverage and division of consistent block on the universe;and according the method,we designed an incre-mental updating algorithm of approximations.The experimental results of six groups of UCI data sets showed that the time of this re-search algorithm was shortened than the non-incremental updating algorithm by 1/3 and 1/2 when adding or deleting objects,respec-tively.In the same data set,the time consumption was less than that of the other four update algorithms.The efficiency and feasibili-ty of this research algorithm was verified.

关 键 词：相容块 划分 覆盖 增量更新 近似集 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]