文献类型：期刊文献

中文题名：基于改进粒子群优化的无标记数据鲁棒聚类算法

英文题名：Improved particle swarm optimization based robust clustering algorithm for unlabeled data

作者：茹蓓[1];朱楠[1];贺新征[2]

第一作者：茹蓓

机构：[1]新乡学院计算机与信息工程学院;[2]河南大学计算机与信息工程学院

第一机构：新乡学院计算机与信息工程学院

年份：2017

卷号：34

期号：6

起止页码：1626-1630

中文期刊名：计算机应用研究

外文期刊名：Application Research of Computers

收录：CSTPCD;;北大核心:【北大核心2014】；CSCD:【CSCD_E2017_2018】；

基金：河南省高等学校青年骨干教师培养计划资助项目(2013GGJS-222);河南省科技厅资助项目(152400410345);河南省科技厅科技攻关项目(172102210445);河南省教育厅资助项目(15A520093)

语种：中文

中文关键词：多目标粒子群优化;聚类算法;鲁棒性;帕累托最优解;无标记数据

外文关键词：multi-objective particle swarm optimization; clustering algorithm; robustness; Pareto optimality; unlabeled data

摘要：已有的聚类算法大多仅考虑单一的目标,导致对某些形状的数据集性能较弱,为此提出一种基于改进粒子群优化的无标记数据鲁棒聚类算法。优化阶段:首先采用多目标粒子群优化的经典形式生成聚类解集合;然后使用K-means算法生成随机分布的初始化种群,并为其分配随机初始化的速度;最终,采用maximin策略确定帕累托最优解。决策阶段:测量帕累托解集与理想解的距离,将距离最短的帕累托解作为最终聚类解。对比实验结果表明,本算法对不同形状的数据集均可获得较优的类簇数量,对目标问题的复杂度具有较好的鲁棒性。
Concerned at the problem that the most existing clustering algorithms only consider single object and they show poor performance in some datasets with particular shapes, this paper proposed an improved PSO based robust clustering algorithm for unlabeled data to resolve above problem. In the optimization phase, firstly, it adopted the classical formation of multi-objective PSO to generate the clustering solution set. Then, it adopted the K-means algorithm to generate the random distributed initial population, and assigned the random initial velocity to each particle. Lastly,it adopted the maximin strategy to decide the Pareto optimality. In the decision phase, it measured the distances between Pareto optimal solutions and ideal solution and selected the shortest one as the final clustering solution. Compared experimental results show that the proposed algorithm show better clustering performance to datasets with different shapes and is robust to the complexity of objective problems.