文章摘要
向培素.一种自适应AP算法的matlab实现[J].西南民族大学自然科学版,2014,40(6):877-882
一种自适应AP算法的matlab实现
The MATLAB program designing of adaptive AP algorithm
中文关键词: 自适应AP算法  Silhouette指标  聚类算法  Matlab
英文关键词: Adaptive AP  Silhouette  Clustering algorithm  Matlab
基金项目:2012年度西南民族大学中央高校基本科研业务费专项项目 ( 12NZYQN05)
作者单位E-mail
向培素 西南民族大学电气信息工程学院 xiangpeisu@swun.cn 
摘要点击次数: 3078
全文下载次数: 1853
中文摘要:
      AP算法是FeyBJ.等人提出的一种聚类算法。与传统的K均值聚类算法相比,AP算法不需要选择初始的聚类中心点,因此,聚类结果更客观。但AP算法中相似度矩阵对角线上的偏向值需要人为设定,而这个值会影响到聚类数目;另外,当AP算法发生震荡时,算法无法自动退出震荡。为解决AP算法中的振荡问题及相似度矩阵对角线上元素值的确定问题,王开军等人提出了自适应AP算法,逐步改变偏向值p,得到不同的聚类结果,再根据聚类结果的Silhouette指标,找出最好的Silhouette指标对应的偏向值及聚类结果。当震荡发生时,逐步增加阻尼因子λ 值,直到算法退出震荡。 使用MATLAB实现了自适应AP算法和Silhouette评价指标,为后续的研究工作打下基础。
英文摘要:
      The AP algorithm is a kind of clustering algorithm proposed by FeyBJ.et al.Compared with the traditional k-means clustering algorithm, the AP algorithm does not need to select the initial exemplare , Therefore, the cluster results are more objective.The diagonal of the similarity matrix in AP algorithm is hard to determine, and the value will affect the clustering number;In addition, when the AP algorithm oscillate algorithm cannot automatically exit.To solve the problem of oscillation of the AP algorithm and to determine the diagonal element value of the similarity matrix, Wang kai-jun et al. proposed adaptive AP algorithm, changing p step by step, obtain the different clustering result, according to the clustering results's Silhouette index, find out the best clustering results.When oscillations occurs, AP algorithm increase the damping factor value step by step, until the oscillation stops.The paper proposed a MATLAB programming of adaptive AP and Silhouette Index.It provides a foundation work for further study.
查看全文   查看/发表评论  下载PDF阅读器
关闭