大规模数据集的多层聚类算法

针对大多谱聚类算法由于计算复杂度高而不适于大规模数据的问题，提出了一种能处理大规模数据集的多层算法。该算法把海量数据根据一定的相关性逐级分组成小数据集，再对分组后的小数据集用谱聚类算法进行聚类，最后利用权核K均值聚类逐级微调，完成全部数据的聚类。通过对UCI数据库中的数据集和图像分割的仿真实验，结果表明该算法的聚类效果很好。

Multilevel Clustering Algorithm for Large Data Sets

This paper introduces a multilevel algorithm for large data sets since most spectral clustering methods are difficult to handle large data sets for its complex computing. A large data sets is divided into small groups level by level according to its intrinsic relevance, then the last subset is clustered by spectral clustering, finally the weighted kernel k - means algorithm is used to perform the refinement of each level. The experiment on the UCI data sets and image segmentation indicate that the algorithm has good results.