基于分组模型的引力搜索智能大数据聚类方法  

Intelligent big data clustering using gravitational search based on grouping

作　　者：胡晓东[1] 高嘉伟[2] HU Xiao-dong;GAO Jia-wei(Department of Electronic Information Engineering,Shanxi Institute of Economic Management,Taiyuan 030024,China;School of Computer and Information Technology,Shanxi University,Taiyuan 030024,China)

机构地区：[1]山西经济管理干部学院电子信息工程系,山西太原030024 [2]山西大学计算机学院,山西太原030024

出　　处：《计算机工程与设计》2021年第6期1660-1667,共8页Computer Engineering and Design

基　　金：山西省自然科学基金项目(2014021022-2)。

摘　　要：提出一种基于分组的引力搜索算法实现数据聚簇。与标准引力搜索不同,分组引力搜索设计一种特定的解编码策略,即分组编码,可将数据聚簇的相关结构映射为解的一部分;对于特定编码,新的引力搜索机制在位置和速度更新策略上设计适合分组编码的更新规则,使分组引力搜索可类似于传统引力搜索进行迭代寻优。在多种经典测试数据集下对算法性能进行评估,其结果表明,与同为智能群体算法的标准引力搜索算法、智能蜂群算法、粒子群算法和萤火虫算法相比,该算法的数据分类效率更高。A gravitational search algorithm based on grouping was proposed to cluster data objects.Different from the standard gravitational search,the grouping gravitational search designed a specific encoding scheme for solutions,called grouping encoding,which was used to make the relevant structures of clustering problems as a part of solutions.Given the specific encoding,the gravitational search mechanism designed a new update rule suitable for the grouping encoding on the location and velocity update,which made the grouping gravitational search algorithm search optimum iteratively as traditional gravitational search algorithm.The performance of the proposed algorithm was evaluated through several classical test datasets.Results show that,compared with same smart and intelligent colony algorithms,such as standard gravitational search algorithm,artificial bee colony algorithm,particle swarm optimization and firefly algorithm,the proposed algorithm has higher efficiency of data clustering.

关 键 词：数据聚簇 分组编码 引力算法 分类失误比率 数据对象距离 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]