基于证据理论的广义多尺度覆盖决策系统的最优尺度组合  

Evidence-Theory-Based Optimal Scale Combinations in Generalized Multi-scale Covering Decision Systems

作　　者：王金波 吴伟志 WANG Jinbo;WU Weizhi(School of Information Engineering,Zhejiang Ocean University,Zhoushan 316022;Key Laboratory of Oceanographic Big Data Mining and Application of Zhejiang Province,Zhejiang Ocean University,Zhou-shan 316022)

机构地区：[1]浙江海洋大学信息工程学院,舟山316022 [2]浙江海洋大学浙江省海洋大数据挖掘与应用重点实验室,舟山316022

出　　处：《模式识别与人工智能》2022年第4期291-305,共15页Pattern Recognition and Artificial Intelligence

基　　金：国家自然科学基金项目(No.61976194,41631179,62076221)资助。

摘　　要：多尺度数据分析是当前粒计算研究领域的热门研究方向,它模拟人类思考模式,以建立多层次的复杂数据和信息处理的有效计算模型为目标.在多尺度数据分析中,一个关键的问题是从系统中选择一个合适的子系统用于最终的分类或决策,这个子系统对应的每个属性的尺度水平的组合称为系统的一个最优尺度组合.针对广义多尺度覆盖决策系统中的知识获取问题,首先,在协调广义多尺度覆盖决策系统中,使用信任函数和似然函数刻画最优尺度组合.然后,在不协调广义多尺度覆盖决策系统中定义7种最优尺度组合的概念,并给出它们之间的关系,结果表明实际有4种不同的最优尺度组合,阐明使用似然函数和信任函数可以定量刻画不协调广义多尺度覆盖决策系统的上、下近似最优尺度组合的特征.最后,举例说明文中方法可用在不完备广义多尺度决策系统和广义多尺度集值决策系统中的最优尺度组合的选择.Multi-scale data analysis is a hot research direction in the field of granular computing.It simulates the mode of human thinking to establish effective computation models for dealing with multi-level complex data and information.A critical problem in multi-scale data analysis is to select a suitable sub-system from a given system for final classification or decision,and the combination of scale level of each attribute corresponding to the sub-system is called an optimal scale combination of the system.To solve the problem of knowledge acquisition in generalized multi-scale covering decision systems,scale combinations are firstly characterized by belief and plausibility functions in consistent generalized multi-scale covering decision systems.Then,the concepts of seven types of optimal scale combinations in inconsistent generalized multi-scale covering decision systems are defined and their relationships are clarified.It is showed that there are actually four different types of optimal scale combinations.Moreover,it is illuminated that belief and plausibility functions can be applied to characterize lower-approximation optimal scale combinations and upper-approximation optimal scale combinations in inconsistent generalized multi-scale covering decision systems,respectively.Finally,it is illustrated that the proposed methods can be applied to the optimal scale combination selection in incomplete generalized multi-scale decision systems and generalized multi-scale set-valued decision systems,respectively.

关 键 词：粒计算 尺度组合 多尺度决策系统 覆盖粗糙集 证据理论 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]