密度峰值自动检测聚类算法  

Clustering Algorithm of Automatic Density Peak Detection

作　　者：吴昊[1] 周建涛[1] 祁瑞东 WU Hao;ZHOU Jiantao;QI Ruidong(College of Computer Science,Inner Mongolia University,Huhhot 010021)

机构地区：[1]内蒙古大学计算机学院,呼和浩特010021

出　　处：《计算机与数字工程》2021年第12期2444-2449,2461,共7页Computer & Digital Engineering

基　　金：国家自然科学基金项目(编号:61662054);内蒙古自然科学基金重大项目(编号:2019ZD15);内蒙古自然科学基金项目(编号:2019GG372);内蒙古自治区高等学校“青年科技英才支持计划”(编号:NJYT-19-A02);内蒙古应用研发项目(编号:201702168)资助。

摘　　要：聚类是一种无监督分类,常用于机器视觉、图像处理等领域。针对密度峰值聚类算法初始聚类中心需要人工手动选择问题,提出一种自动选择聚类中心算法,通过使用最小二乘法结合决策图进行簇中心选择,并提出改进选点策略进一步减少算法所需时间开销。实验中使用Matlab实现算法,并使用6种UCI常用数据集测试,实验结果表明它的性能优于现有的方法,在无需人为干预的情况下获得较好的聚类效果,对数据集内部规律和噪声点都有较好的识别能力。Clustering is a kind of unsupervised classification,which classifies data according to the similarity between data ob⁃jects.It is often used in machine vision and image processing.Aiming at the problem that the initial cluster center of density peak clustering algorithm needs to be manually selected,an automatic cluster center selection algorithm is proposed,which uses the least square method to construct a decision graph to select the cluster center,and proposes an improved point selection strategy to further reduce the time consumption of the algorithm.In the experiment,Matlab is used to implement the algorithm,and six UCI data sets are used to test the algorithm.The experimental results show that its performance is better than the existing methods,and it can ob⁃tain better clustering effect without human intervention,and it has better recognition ability for the internal rules and noise points of the data set.

关 键 词：聚类分析 簇中心 密度峰值 最小二乘法 自动聚类 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]