改进的自适应谱聚类NJW算法
发布时间：2018-09-05 11:36
【摘要】：聚类算法是近年来国际上机器学习领域的一个新的研究热点。为了能在任意形状的样本空间上聚类,学者们提出了谱聚类和图论聚类等优秀的算法。首先介绍了图论聚类算法中的谱聚类经典NJW算法和NeiMu图论聚类算法的基本思路,提出了改进的自适应谱聚类NJW算法。提出的自适应NJW算法的优点在于无需调试参数,即可自动求出聚类个数,克服了经典NJW算法需要事先设置聚类个数且需反复调试参数δ才能得出数据分类结果的缺点。在UCI标准数据集及实测数据集上对自适应NJW算法与经典NJW算法、自适应NJW算法与NeiMu图论聚类算法进行了比较。实验结果表明,自适应NJW算法方便快捷,且具有较好的实用性。
[Abstract]:Clustering algorithm is a new research hotspot in the field of machine learning in the world in recent years. In order to cluster on arbitrary shape sample space, scholars have proposed spectral clustering and graph theory clustering algorithms. Firstly, the basic ideas of spectral clustering classical NJW algorithm and NeiMu graph theory clustering algorithm in graph theory clustering algorithm are introduced. An improved adaptive spectral clustering NJW algorithm is proposed. The advantage of the proposed adaptive NJW algorithm is that the number of clusters can be automatically calculated without debugging parameters. The disadvantage of the classical NJW algorithm is that the number of clusters needs to be set in advance and the parameter delta needs to be debugged repeatedly to get the classification results. The adaptive NJW algorithm is compared with the classical NJW algorithm and the adaptive NJW algorithm is compared with the NeiMu graph theory clustering algorithm.
【作者单位】： 大连理工大学计算机科学与技术学院;
【基金】：国家大学生创新创业训练项目(2016101410168)资助
【分类号】：O157.5;TP311.13
本文编号：2224150
[Abstract]:Clustering algorithm is a new research hotspot in the field of machine learning in the world in recent years. In order to cluster on arbitrary shape sample space, scholars have proposed spectral clustering and graph theory clustering algorithms. Firstly, the basic ideas of spectral clustering classical NJW algorithm and NeiMu graph theory clustering algorithm in graph theory clustering algorithm are introduced. An improved adaptive spectral clustering NJW algorithm is proposed. The advantage of the proposed adaptive NJW algorithm is that the number of clusters can be automatically calculated without debugging parameters. The disadvantage of the classical NJW algorithm is that the number of clusters needs to be set in advance and the parameter delta needs to be debugged repeatedly to get the classification results. The adaptive NJW algorithm is compared with the classical NJW algorithm and the adaptive NJW algorithm is compared with the NeiMu graph theory clustering algorithm.
【作者单位】： 大连理工大学计算机科学与技术学院;
【基金】：国家大学生创新创业训练项目(2016101410168)资助
【分类号】：O157.5;TP311.13
【相似文献】
相关期刊论文 前3条
1 徐步东;;基于谱聚类的社团发现算法[J];潍坊工程职业学院学报;2013年04期
2 张振宇;张珍;杨文忠;吴晓红;;复杂网络中重叠社区检测[J];计算机工程与科学;2013年12期
3 ;[J];;年期
相关硕士学位论文 前3条
1 陈昭彤;基于拓扑势和谱聚类的社区发现研究[D];中国矿业大学;2015年
2 郭敏;基于谱聚类的复杂网络社团结构发现算法[D];西安电子科技大学;2014年
3 杜鹏;基于谱聚类和CHSMM的非线性系统的剩余寿命预测[D];哈尔滨工业大学;2011年
，本文编号：2224150
本文链接：https://www.wllwen.com/kejilunwen/yysx/2224150.html
