不完备程度多粒度粗糙集模型研究

发布时间：2018-05-24 01:18

本文选题：限制容差关系 + 程度多粒度粗糙集　；参考：《安徽工业大学》2017年硕士论文

【摘要】：由于网络技术和通信技术的飞速发展,涌现出类型多样的海量数据,各个领域都期待从海量的、杂乱无章的、噪声数据中获取有用的知识。粗糙集(rough set,RS)在获取模糊性、不确定的知识方面展现出巨大的优势。它不需要任何其它先验知识和附加信息,依靠数据集合本身的属性,便可以挖掘出数据中隐含的有价值的信息。多粒度粗糙集(Multi-Granulation Rough Set,MGRS)是一种新的粗糙集扩展模型,它从多个粒度空间对目标概念进行近似逼近,在边界区域的范围缩小,目标概念的表示精度提高方面,具有明显的优势。实际生活中,由于测量偏差等因素,常常存在一些不完备的,但隐藏着丰富知识的数据。为了从不完备信息系统中获得更加准确的知识,本文结合程度粗糙集,研究不完备MGRS模型和粒度约简方法。本文主要工作如下:(1)介绍经典粗糙集的基础知识,给出一些实例形象地解释粗糙集的基本概念。针对完备系统和不完备系统,介绍目前MGRS的发展与研究现状。(2)分别介绍了基于容差关系、相似关系、限制容差关系的单粒度粗糙集拓展模型和MGRS拓展模型,分析不同关系下各个粗糙集模型的优缺点。(3)针对不完备信息系统,提出基于限制容差关系的程度MGRS,包括程度乐观MGRS和程度悲观MGRS。分析程度乐观MGRS和程度悲观MGRS的不足之处,提出一种基于限制容差关系的可变程度MGRS模型。研究这三种模型的相关性质与联系。通过实例和实验分析可变程度MGRS的优越性。(4)考虑粒度的权重,基于限制容差关系,提出不完备加权程度MGRS,并讨论其性质。定义不完备加权程度MGRS的粒度矩阵、核粒度和粒度重要性公式。提出一种粒度约简方法,在获取核粒度的基础上,以粒度重要性作为启发式信息选择粒度,获得最终的粒度约简集。
[Abstract]:Due to the rapid development of network technology and communication technology, massive data of various types have emerged. All fields expect to obtain useful knowledge from mass, disorderly and noisy data. Rough set sets (RS) show great advantages in acquiring fuzzy and uncertain knowledge. It does not need any other prior knowledge and additional information. It can mine the valuable information hidden in the data by relying on the attributes of the data set itself. Multi-granulation Rough set (MGRS) is a new rough set extension model, which approximates the concept of target from multiple granularity spaces, and has obvious advantages in reducing the range of boundary area and improving the precision of representation of target concept. In real life, because of the measurement deviation and other factors, there are often some incomplete, but hidden knowledge of the data. In order to obtain more accurate knowledge from incomplete information system, this paper studies incomplete MGRS model and granularity reduction method combining degree rough set. The main work of this paper is as follows: (1) introduce the basic knowledge of classical rough sets and give some examples to explain the basic concepts of rough sets graphically. For complete systems and incomplete systems, the development and research status of MGRS are introduced. (2) the single granularity rough set extension model and MGRS extension model based on tolerance relation, similarity relation and limiting tolerance relation are introduced respectively. This paper analyzes the advantages and disadvantages of each rough set model under different relationships. (3) aiming at incomplete information systems, the degree MGRS based on restricted tolerance relationship is proposed, including degree optimistic MGRS and degree pessimistic MGRs. By analyzing the disadvantages of degree optimistic MGRS and degree pessimistic MGRS, a variable degree MGRS model based on restricted tolerance relationship is proposed. The related properties and relationships of these three models are studied. The advantage of variable degree MGRS is analyzed by examples and experiments. Considering the weight of granularity, based on the limited tolerance relation, the incomplete weighted degree MGRS is proposed and its properties are discussed. The granularity matrix, kernel granularity and granularity importance formula of incomplete weighted degree MGRS are defined. A granularity reduction method is proposed. Based on the kernel granularity, granularity importance is used as heuristic information to select granularity, and the final granularity reduction set is obtained.
【学位授予单位】：安徽工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP18

【参考文献】