一种改进的局部多粒度决策理论粗糙集模型  

An Improved Local Multi-granulation Decision-theoretic Rough Set Model

作　　者：张鹏飞[1,2] 李天瑞[1,2] 王德贤 袁钟 王国强[1,2] ZHANG Peng-fei;LI Tian-rui;WANG De-xian;YUAN Zhong;WANG Guo-qiang(School of Computing and Artificial Intelligence,Southwest Jiaotong University,Chengdu 611756,China;National Engineering Laboratory of Integrated Transportation Big Data Application Technology(Southwest Jiaotong University),Chengdu 611756,China)

机构地区：[1]西南交通大学计算机与人工智能学院,四川成都611756 [2]综合交通大数据应用技术国家工程实验室(西南交通大学),四川成都611756

出　　处：《模糊系统与数学》2022年第6期40-53,共14页Fuzzy Systems and Mathematics

基　　金：国家自然科学基金资助项目(62176221,61573292)。

摘　　要：局部多粒度决策理论粗糙集要预先获取给定数据集中所有对象的信息颗粒,只需要对特定的目标概念中的对象的信息颗粒进行计算,开创了一种有用的计算范式。然而,传统的局部多粒度决策理论粗糙集在计算三个区域(正域,边界域和负域)时需要主观的给定一对概率阈值(α,β)。在实际的决策应用中,该获取阈值的方法可能会造成信息丢失或判断不准确的问题。为了解决这个问题,这篇文章提出了一种改进的局部多粒度决策理论粗糙集模型,叫做广义的局部多粒度决策理论粗糙集。该模型可以通过一个补偿系数ζ,即可自适应的获得相对应的参数α和β.这不仅减少了人为设置参数的个数,还强化了由多个粒度结构所产生损失的语义解释。The local multi-granulation decision-theoretic rough set does need not to obtain information granules for all the objects in a given data set in advance, but instead it only computes them for objects from a specific target concept, which develops a useful computing paradigm. However, the traditional local multi-granulation decision-theoretic rough sets require subjectively given a pair of probability thresholds(α,β) when calculating three regions(positive region, boundary region, and negative region). In practical decision-making applications, this method of obtaining thresholds may lead to loss of information or inaccurate judgments. To solve this problem, this paper proposes an improved model of local multi-granulation decision-theoretic rough set called a generalized local multi-granulation decision-theoretic rough set. This model can adaptively obtain the corresponding parameters α and β via a compensation coefficient ζ. This not only minimizes the amount of parameters that have been artificially established, but also enhances the semantic interpretation of losses from multiple granularity structures.

关 键 词：粗糙集 局部多粒度决策理论粗糙集 自适应 补偿系数 粒计算 

分 类 号：O159[理学—数学]