噪声环境下复杂流形数据的势能层次聚类算法

基于势能的快速凝聚层次聚类算法使用一种全新的相似性度量准则，可以更高效地得到聚类结果。针对该算法无法有效处理含噪声的复杂流形数据的缺陷，提出噪声环境下复杂流形数据的势能层次聚类算法。通过势能递增曲线识别噪声点，在新定义的势能最大、最小2层数据上进行自动聚类，以确定类簇的大体框架，并在此基础上对整个数据集进行层次聚类。人工数据集上的实验表明，新算法可以有效处理噪声环境下复杂流形数据；真实数据集上的实验表明，新算法具有更优的聚类效果。

Hierarchical clustering algorithm based on potential in complex flow data sets with noise

Potential-based hierarchical agglomerative (PHA) clustering uses a new similarity metric to get clustering results more efficiently. Aiming at the problem that PHA can not effectively deal with the complex structure data sets with noise, we proposed a hierarchical clustering algorithm based on potential in complex flow data sets with noise. Firstly, the noise points were identified by means of the potential increase curve; Secondly, the maximum and minimum layers of the data set based on potential were defined, and these two data layers were made to cluster automatically to determine the general framework of the clusters; Finally, clustering of the entire data set was hierarchied based on the precious clustering results. The experiments on artificial data sets show that the new algorithm can deal with the complex flow data sets with noise effectively. The experiments on real data sets show that the new algorithm can get the better clustering results.