利用多标签信息的聚类方法

One clustering method by using multi-label information

k-means是当前使用最广泛的聚类算法,但其只对无标签信息的样本数据进行聚类,需用户提供初始聚类中心等信息且聚类花费时间长结果不稳定。当前很多数据集都带有多标签的信息,如何利用多标签信息更加快速有效的聚类是一个重要的研究方向。文中提出了多标签数据的聚类方法,首先对多标签信息进行聚类得到适当的初始聚类中心,然后再使用k-means算法对大量样本做聚类。试验结果表明,该聚类方法比使用随机初始聚类中心的k-means算法具有更好的时间性能和聚类结果。

K-means algorithm is currently most widely used clustering method,but it only clusters unlabeled sample data,which requires user to provide initial clusters centers and achieve unstable results at long time.Currently,many real datasets contain both samples and associated multi-label information.It is an important research topic how to do clustering with multi-label information rapidly and effectively.The article proposes a clustering method for multi-label information,which we do clustering on multi-label vectors to acquire appropriate initial clusters centers firstly and then use the k-means algorithm to cluster a large number of samples.The experimental results illustrate that the proposed method achieves better time performance and clustering results than the k-means algorithm using random initial clustering centers.