多粒度标记决策系统中最优粒度的选择

发布时间：2018-04-14 16:23

本文选题：粒计算 + 信息系统　；参考：《浙江海洋大学》2017年硕士论文

【摘要】：粗糙集理论是波兰数学家Pawlak在1982年提出的一种数学理论,它可以处理分析不完备数据,其主要的特点是在不改变分类能力的条件下,通过剔除冗余的信息来获得知识的属性约简,从而进行规则提取。粗糙集数据分析的特点是不需要提供所给数据外的任何假设,只利用所给的数据做出决策。随着粗糙集理论的研究与发展,现已在很多领域成功得到应用,近年来在农业信息化领域中也获得了较多成果。粒计算是数据挖掘和知识表示的一种重要方法,是当前人工智能,特别是智能信息处理领域非常活跃的研究方向,它以粒为计算单位,模拟人的思考模式,通过分析和处理大规模复杂的数据和建立有效的计算模型为目标。粒计算的特点是在于寻求在某个合适的粒度层面下用近似方法来解决复杂问题。在现实生活中,人们常从不同粒度层面去观察对象或处理数据,在不同粒度层面下选择合适的决策系统和对应的决策规则是智能信息处理中的一个非常重要的研究课题。针对具有多粒度标记决策系统的知识获取问题,本文以粒计算为基础研究在多粒度标记决策系统中的最优粒度选择。首先,介绍了信息系统、多粒度标记信息系统、以及多粒度标记决策系统等概念。其次,讨论多粒度标记决策系统在不同意义下的最优粒度选择,介绍了多粒度标记决策系统中的8种协调性和最优粒度的概念,并讨论了它们之间的相互关系,证明了实际上只有4种不同类型的最优粒度。最后,在多粒度标记决策系统中分别给出了基于4种最优粒度的属性约简与规则提取方法。
[Abstract]:Rough set theory is a kind of mathematical theory put forward by Polish mathematician Pawlak in 1982. It can deal with and analyze incomplete data. Its main characteristic is that the classification ability is not changed.The attribute reduction of knowledge is obtained by eliminating redundant information, and the rules are extracted.The feature of rough set data analysis is that it does not need to provide any assumptions outside the given data, and only makes decisions using the given data.With the research and development of rough set theory, it has been successfully applied in many fields. In recent years, many achievements have been obtained in the field of agricultural informatization.Granular computing is an important method of data mining and knowledge representation. It is a very active research direction in the field of artificial intelligence, especially intelligent information processing.The objective is to analyze and process large-scale and complex data and to establish effective computing models.The characteristic of granular computing is to find an approximate method to solve complex problems at a suitable granularity level.In real life, people often observe objects or deal with data from different granularity levels. Choosing appropriate decision system and corresponding decision rules at different granularity levels is a very important research topic in intelligent information processing.Aiming at the knowledge acquisition problem of multi-granularity label decision system, this paper studies the optimal granularity selection in multi-granularity label decision system based on granular computing.Firstly, the concepts of information system, multi-granularity label information system and multi-granularity label decision system are introduced.Secondly, the optimal granularity selection of multi-granularity label decision system in different sense is discussed. Eight concepts of coordination and optimal granularity in multi-granularity label decision system are introduced, and the relationship between them is discussed.It is proved that there are only four different types of optimal granularity.Finally, four methods of attribute reduction and rule extraction based on optimal granularity are presented in the multi-granularity label decision system.
【学位授予单位】：浙江海洋大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP18

【相似文献】