基于增强重引力搜索的高维数据协同聚类算法  

CO-CLUSTERING ALGORITHM FOR HIGH DIMENSIONAL DATA BASED ON ENHANCED GRAVITY SEARCH ALGORITHM

作　　者：桑遥 尹君 王迪 王皓 景康 Sang Yao;Yin Jun;Wang Di;Wang Hao;Jing Kang(Urumqi Power Supply Company,State Grid Xinjiang Electric Power Corporation,Urumqi 830002,Xinjiang,China;Information and Communication Corporation,State Grid Xinjiang Electric Power Corporation,Urumqi 830018,Xinjiang,China)

机构地区：[1]国网新疆电力有限公司乌鲁木齐供电公司,新疆乌鲁木齐830002 [2]新疆电力有限公司信息通信公司,新疆乌鲁木齐830018

出　　处：《计算机应用与软件》2020年第10期300-306,共7页Computer Applications and Software

摘　　要：传统基于智能优化技术的聚类算法难以获得理想的聚类准确率。设计一种基于增强重引力搜索的高维数据协同聚类算法,用以提高高维数据的聚类准确率。设计协同相似性度量技术同时评估样本的相似性和特征的相似性,以特征间的相似性引导数据样本的聚类处理,提高聚类的效率和准确率。设计拟牛顿法的局部开发机制,提高重引力搜索的求解效果,利用增强的重引力搜索算法搜索最优的聚类解。实验结果表明,该方法对高维数据集的聚类结果具有明显的优势。Traditional clustering algorithms based on the intelligence optimization technologies are difficult to obtain ideal clustering accuracy.We propose a co-clustering algorithm for high dimensional data based on enhanced gravity search algorithm to improve the clustering accuracy for high dimensional data.We designed a co-similarity measure technique to evaluate the similarities of samples and features at the same time.The similarity between features guided the clustering processing of data samples to improve the efficiency and accuracy of clustering.The local development mechanism of quasi-Newton method was designed to improve the solution quality of gravity search,and the enhanced gravity search algorithm was used to search the optimal clustering solution.The experimental results show that our method has obvious advantages for clustering results of high dimensional datasets.

关 键 词：协同聚类 高维数据 特征选择 重引力搜索 数据挖掘 拟牛顿法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]