基于对象更新的邻域多粒度粗糙集模型增量式算法

An incremental algorithm for the neighborhood multi-granulation rough set model based on object update

邻域多粒度粗糙集模型是粗糙集理论的重要研究分支。然而在大数据环境下,数据时刻处于动态更新之中,针对数值型信息系统对象动态变化的情形,本文提出一种邻域多粒度粗糙集模型的增量式更新算法。文中首先利用矩阵的方法表示了邻域多粒度粗糙集中邻域类与目标近似集之间的两种近似关系,分别称之为子集近似关系矩阵和交集近似关系矩阵,并通过这两种近似关系矩阵重构了邻域多粒度粗糙集模型;然后针对数值型信息系统对象增加和对象减少的情形,研究了这两种近似关系矩阵随对象变化时的增量式更新,理论分析证明了这种更新方法的高效性;最后基于近似关系矩阵的增量式更新设计出了邻域多粒度粗糙集模型的增量式更新算法。实验结果验证了所提出增量式算法的有效性和优越性。

The neighborhood multi-granulation rough set model is an important branch of rough set theory.However,in the big data environment,the data is constantly being updated dynamically.In view of the dynamic changes of numerical information system objects,an incremental updating algorithm for neighborhood multi-granulation rough set model is proposed in this paper.Firstly,two kinds of approximation relations between neighborhood class and target approximation set in the neighborhood multi-granulation rough sets are expressed by the matrix method,which are called subset approximation relation matrix and intersection approximation relation matrix,respectively.The neighborhood multi-granulation rough set model is reconstructed by these two approximation relation matrices.Then,the incremental updating of these two approximation relation matrices is studied in the case of increasing and decreasing objects in numerical information system.The theoretical analysis proves that this updating method is of high-efficiency.Finally,the incremental updating algorithm of neighborhood multi-granulation rough set model is designed based on the incremental updating of approximation relation matrices.Experimental results verify the effectiveness and superiority of the proposed incremental algorithm.