不完备信息系统中基于粗糙集的规则提取研究

发布时间：2018-04-13 11:20

本文选题：粗糙集 + 不完备信息系统　；参考：《浙江大学》2017年硕士论文

【摘要】：随着计算机和网络信息技术快速发展,人们获取的数据急剧膨胀,对数据进行智能分析变得特别重要。特别地,现实中的数据往往存在着噪音和缺失,处理数据的不确定性和不完备性显得尤为重要。粗糙集理论是由波兰学者Z.Pawlak提出的一种有效刻画不确定性的数学工具。目前,粗糙集理论已经在多个领域得到广泛应用,其中,规则提取是最为重要的应用之一。目前已有一些基于粗糙集理论的规则提取算法,其中DRICA算法从属性空间的观点提取规则,而且不需要提前进行属性约简,具有一定的优势。但其不能直接处理不完备信息系统,而不完备信息系统在现实中广泛存在。基于DRICA,本文提出了两种新的规则提取算法,继承了 DRICA算法的优点,并且可以直接处理不完备信息系统。在不完备信息系统中上下近似算子有三种不同的定义,本文通过实验比较了不同近似定义对提出的算法的影响。经典粗糙集的上、下近似过于严格,限制了粗糙集的应用,本文进一步将提出的规则提取算法扩展到了变精度粗糙集。具体来说,本文的主要工作如下:·基于DRICA,提出了两种新的适用于不完备信息系统的规则提取算法,并通实例分析、实验分析和与LEM2算法的实验对比,显示了算法的有效性。·通过实验,比较了不同的上近似定义对所提出的两种规则提取算法的影响。·将本文中提出的两种规则提取算法扩展到了变精度粗糙集模型,通过实验对不同的变精度系数进行了比较,并对变精度系数的取值给出了一些建议。
[Abstract]:With the rapid development of computer and network information technology, the data obtained by people expand rapidly, so it is very important to analyze the data intelligently.In particular, there are always noises and defects in the real data, so it is very important to deal with the uncertainty and incompleteness of the data.Rough set theory is an effective mathematical tool for characterizing uncertainty proposed by Polish scholar Z.Pawlak.At present, rough set theory has been widely used in many fields, among which rule extraction is one of the most important applications.At present, there are some rules extraction algorithms based on rough set theory, in which DRICA algorithm extracts rules from the viewpoint of attribute space and does not need attribute reduction in advance, so it has some advantages.But it can not deal with incomplete information system directly, and incomplete information system exists widely in reality.In this paper, two new rules extraction algorithms based on DRICA are proposed, which inherit the advantages of DRICA algorithm and can deal with incomplete information systems directly.There are three different definitions of upper and lower approximation operators in incomplete information systems.The upper and lower approximations of classical rough sets are too strict, which limits the application of rough sets. In this paper, the proposed rule extraction algorithm is further extended to variable precision rough sets.Specifically, the main work of this paper is as follows: based on rica, two new rules extraction algorithms for incomplete information systems are proposed, and an example analysis, experimental analysis and experimental comparison with LEM2 algorithm are given.The effectiveness of the algorithm is demonstrated. Through experiments, the effects of different upper approximation definitions on the proposed two rules extraction algorithms are compared, and the two rules extraction algorithms proposed in this paper are extended to the variable precision rough set model.The different variable precision coefficients are compared by experiments, and some suggestions are given for the value of the variable precision coefficients.
【学位授予单位】：浙江大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP18

【相似文献】