Search - CURE data clustering algorithm

Search - CURE data clustering algorithm - List

[Mathimatics-Numerical algorithms] cure(Clustering) DL : 0: CURE（Clustering Using Representatives）是一种针对大型数据库的高效的聚类算法。基于划分的传统的聚类算法得到的是球状的，相等大小的聚类，对异常数据比较脆弱。CURE采用了用多个点代表一个簇的方法，可以较好的处理以上问题。并且在处理大数据量的时候采用了随机取样，分区的方法，来提高其效率，使得其可以高效的处理大量数据。-CURE (Clustering Using Representatives) is a database for large and efficient clustering algorithm. Based on the breakdown of the traditional clustering algorithm is spherical, equal-sized cluster of abnormal data more vulnerable. CURE adopted by a number of points on behalf of a cluster approach can better deal with these questions. And deal with large amount of data when using the random sampling, area way to improve its efficiency so that it can handle large amounts of data and efficient.
Date : 2025-12-21 Size : 21kb User : 肖宪
[Mathimatics-Numerical algorithms] CureDemo DL : 0: 实现的cure聚类的demo。算法在开始时，每个点都是一个簇，然后将距离最近的簇结合，一直到簇的个数为要求的K。它是一种分裂的层次聚类。算法分为以下6步： 1）从源数据对象中抽取一个随机样本S。 2）将样本S分割为一组划分。 3）对划分局部的聚类。 4）通过随机取样提出孤立点。如果一个簇增长得太慢，就去掉它。 5）对局部的簇进行聚类。 6）用相应的簇标签标记数据。(The implementation of the cure clustering of the demo. At the beginning of the algorithm, each point is a cluster, and then the nearest cluster is combined to the number of clusters K. It is a hierarchical cluster of divisions. The algorithm is divided into the following 6 steps: 1) extract a random sample S from the source data object. 2) the sample S is divided into a set of partitions. 3) to divide the local clustering. 4) the isolated point is proposed by random sampling. If a cluster is growing too slowly, get rid of it. 5) clustering the local clusters. 6) mark the data with the corresponding cluster tag.)
Date : 2025-12-21 Size : 158kb User : Aileen00080