高维数据多级模糊模式识别的分类研究

Classification research based on multiple fuzzy pattern recognition in high-dimensional data

通过分析对象属性间的关系,提出了一种基于改进的多级模糊模式识别的分类方法。该方法重点考虑对象属性间影响较大的因素,以此建立影响对象分类的属性之间的简化关系,使分类结果更加合理;针对分类标准为对象属性分类的离散值,存在对象属性值介于中间状态不便分类问题,通过建立属性值所属级别的矩阵来确定属性权重,使分类精确;利用Rough集的特征属性约简算法降低数据集的维数,提高高维数据的分类效率。经实例证明该方法分类准确、效率高。

By analysing the relation among object attribute, this paper introduced a classification method based on improved multiple fuzzy pattern recognition. This method mainly took the factors into account which were greatly affected by object attribute, so as to find out the concise relation affecting object classification among attribute and make the classification result more rational; as for the question that it was inconvenient for classification owing to betweenness because classification standard was described by discrete value of object cassification, introduced a matrix of attribute value belonging to classification to estimate attribute weight, so it could make the classification more accurate. This study made use of algorithm on the characteristic attribute reduction of rough set in order to reduce dimension of data set and enhance classification efficiency in high-dimensional data. It proves this method makes classification more accurate and more efficient with the example.