优势关系下基于论域递减的属性约简方法

发布时间：2018-02-10 11:45

本文关键词： 粗糙集优势类正域辨识矩阵快速约简　出处：《河北大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着科学技术的进步与发展,每天产生数以万计的数据,如何从这些数据中挖掘出重要的、有用的信息显得尤为重要。在实际问题中,描述对象的属性值大多是具有偏好顺序的,而粗糙集理论的核心问题一直是属性约简及其效率的提高,因此优势关系下的快速属性约简方法的研究是非常有意义的。在优势关系下粗糙集模型的属性约简方法中,计算优势类作为基本粒进而计算上下近似集是必不可少的步骤,而传统算法需要比较所有样例的每个属性值,耗时较大,近似集、属性约简的计算效率受到了影响。大部分已有的加速算法都是针对等价类的计算,只能用于处理符号值属性。基于此,本文根据优势类的性质提出了一种快速缩减搜索空间的思想,并在此基础上设计相应的算法来提高优势类的计算效率,从而提高属性约简的性能,尤其适用于属性数量较多的情况。进一步,在快速求得优势类的基础上将优势类和决策类表示为01矩阵,采用或运算计算正域,利用辨识矩阵求得属性约简。最后通过UCI数据集的实验表明,该方法在与传统方法保持相同约简集的情况下,减少了计算时间,提高了属性约简的效率。
[Abstract]:With the progress and development of science and technology, tens of thousands of data are produced every day. It is very important to extract important and useful information from these data. Most of the attribute values of describing objects have preference order, but the core problem of rough set theory is attribute reduction and the improvement of its efficiency. Therefore, it is very meaningful to study the fast attribute reduction method in the dominant relation. In the attribute reduction method of rough set model under the advantage relation, it is necessary to calculate the superior class as the basic particle and then calculate the upper and lower approximate set. However, the traditional algorithm needs to compare each attribute value of all the samples, which is time-consuming, approximate set, and the efficiency of attribute reduction is affected. It can only be used to deal with symbolic value attributes. Based on this, this paper puts forward an idea of reducing search space quickly according to the properties of superior classes, and then designs corresponding algorithms to improve the computational efficiency of superior classes. In order to improve the performance of attribute reduction, especially in the case of a large number of attributes. Further, the dominant class and decision class are expressed as 01 matrix on the basis of fast finding the superior class, and the positive domain is calculated by or by calculation. Finally, the experiment of UCI data set shows that the method can reduce the computation time and improve the efficiency of attribute reduction by keeping the same reduction set as the traditional method.
【学位授予单位】：河北大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP18

【参考文献】