摘要
粒计算是知识表示与数据处理的一种新的理念和计算范式,现已成为复杂问题求解的有效框架之一。信息粒是粒计算的基本单位,构建信息粒以及利用信息粒实现问题求解是粒计算的两项基本任务。受认知主体的主观局限性和度量方法等因素的影响,信息粒的形成、表示与解释往往伴随着不确定性。本文基于区间二型模糊C均值(IT2FCM)聚类算法和合理粒度原则设计了二型信息粒的生成方法,不仅实现了信息粒的构建,而且能够体现出粒化过程中的不确定性。为验证所设计方法的合理性,本文在人工数据集上对粒化方法进行评估,并探讨不同参数值对粒化结果的影响。此外,利用所设计的粒化方法对船舶数据的结构信息及其不确定性进行刻画,以此来辅助船舶的分类管理。
Granular computing is an emerging principle and computing paradigm in representing knowledge and processing dataset,and is one of the effective frameworks to solve complex problems.As the fundamental elements,the construction of information granules and how to use it in problem solving are two basic issues in granular computing.Influenced by the subjective limitations of cognitive subjects and measurement methods,the formation,presentation,and interpretation of information granules are often accompanied by uncertainty.In this paper,with the aid of the type-2 Fuzzy C-Means(IT2 FCM)algorithm and principle of justifiable granularity,a new strategy for generating type-2 information granules is designed,which not only realizes the construction of information granules,but also reflects the uncertainty in granulation process.To verify the rationality of the designed method,this paper evaluates the performance of the granulation model for the synthetic datasets,and examines the impacts of different parameter values on the granulation results.In addition,the designed method is employed to the ship dataset for assisting the ship management through the capture of data structure and its uncertainty.
作者
赵芳
郭红月
王利东
ZHAO Fang;GUO Hong-yue;WANG Li-dong(School of Science,Dalian Maritime Univervsity,Dalian 116026,China;School of Maritime Economics and Management,Dalian Maritime University,Dalian 116026,China)
出处
《模糊系统与数学》
北大核心
2021年第1期101-110,共10页
Fuzzy Systems and Mathematics
基金
国家自然科学基金资助项目(61773352,62006033)
中央高校基本科研业务费专项(3132019357)。
关键词
信息粒
合理粒度原则
模糊聚类
区间二型模糊C均值聚类
Information Granule
Principle of Justifiable Granularity
Fuzzy Clustering
Interval Type-2 Fuzzy C-Means Clustering
