基于局部线性嵌入的半监督仿射传播聚类算法  

Semi-supervised affinity propagation clustering algorithm based on local linear embedding

作　　者：赵小强[1] 谢亚萍[1] 

机构地区：[1]兰州理工大学电气工程与信息工程学院,甘肃兰州730050

出　　处：《兰州理工大学学报》2015年第1期96-100,共5页Journal of Lanzhou University of Technology

基　　金：国家自然科学基金(51265032;61263003);甘肃省高校基本科研业务费项目(1203ZTC061)

摘　　要：针对运用半监督仿射传播聚类算法处理高维数据时聚类精度低和计算量大的问题,提出一种基于局部线性嵌入的半监督仿射传播聚类算法.该算法首先通过LLE算法将高维输入数据集映射到低维空间得到低维数据集,计算低维数据集的相似度矩阵,再用半监督算法调整相似度矩阵,最后用仿射传播聚类算法对低维数据进行聚类分析.仿真结果表明,本文提出的算法与半监督仿射传播聚类算法相比,在处理高维数据时聚类效果更好,精度更高,迭代次数更少.Aimed at the problem of low clustering precision and time-consuming calculation of high-dimensional data sets with semi-supervised affinity propagation clustering algorithm,a local linear embeddingbased clustering algorithm of semi-supervised affinity propagation（LLE-SAP）is proposed.Firstly,with this algorithm,the high-dimensional input data set will be mapped into low-dimensional space with LLE algorithm to get a low-dimensional data set.Then the similarity matrix of the low-dimensional data is evaluated and adjusted with semi-supervised algorithm.Finally,the clustering analysis of the low-dimensional data will be conducted with affinity propagation clustering algorithm.Simulation result shows that compared with the algorithm available,the proposed algorithm will have higher precision,fewer iterations,and better effect for treating high-dimensional data.

关 键 词：数据挖掘 半监督 仿射传播聚类 局部线性嵌入算法 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]