局部加权邻域多粒度粗糙集  

Local Weighted Neighborhood-based Multigranulation Rough Sets

作　　者：樊祥宁 张燕兰 FAN Xiang-ning;ZHAMG Yan-lan(School of Computer Science,Minnan Normal University,Zhangzhou 363000,China;Key Laboratory of Data Science and Intelligence Application,Zhangzhou 363000,China)

机构地区：[1]闽南师范大学计算机学院,福建漳州363000 [2]数据科学与智能应用福建省高校重点实验室,福建漳州363000

出　　处：《模糊系统与数学》2022年第6期64-78,共15页Fuzzy Systems and Mathematics

基　　金：国家自然科学基金资助项目(11701258,11871259);福建省自然科学基金资助项目(2020J01801,2020J02043,2019J01749);福建省高校杰出青年科研人才培养计划项目。

摘　　要：为处理粒度质量不均衡型数据,加权多粒度粗糙集及加权平均多粒度决策粗糙集被先后提出,其拓宽了多粒度粗糙集的应用范围。然而,随着数据规模剧增,传统模型已无法满足实际需求。利用矩阵算法计算近似算子有利于提高计算效率,但其空间复杂度相对较高。为此,本文提出局部加权邻域多粒度粗糙集模型,将局部粗糙集模型与矩阵理论相结合以降低矩阵算法的时间和空间复杂度。首先,给出局部加权邻域多粒度粗糙集模型的定义和性质;随后,设计出计算近似算子的矩阵算法。最后,通过实验在6个UCI数据集中验证局部算法比全局算法具有更高的时间效率。To handle data with unbalanced granularity and quality, multigranulation rough sets based on weighted granulations and weighted average multigranulation decision rough sets have been proposed, which broaden the application scope of multigranulation rough sets. However, with the rapid increase of data scale, the traditional models can no longer meet the practical requirements. Using the matrix algorithm to calculate the approximation operator is beneficial to improve computational efficiency, but its space complexity is relatively high. Therefore, a local weighted neighborhood-based multigranulation rough sets model is proposed in this paper, which combines the local rough set model with matrix theory to reduce the time and space complexity of the algorithm by matrix only. Firstly, the definition and properties of the local weighted neighborhood-based multigranulation rough sets model are given. Then, a matrix algorithm for calculating the approximation operator is designed to reduce the computational redundancy. Finally, experiments show that the local algorithm has higher time efficiency than the global algorithm in six UCI datasets.

关 键 词：局部粗糙集 加权多粒度粗糙集 邻域信息系统 矩阵 近似算子 

分 类 号：O159[理学—数学]