一种基于在线凝聚的层次聚类改进算法

AN IMPROVED HIERARCHICAL CLUSTERING ALGORITHM BASED ON ONLINE AGGLOMERATION

为了快速有效地聚类增量式数据,针对传统在线凝聚聚类算法扁平结构的不足以及执行速度的劣势,在Add C和AHOC算法基础上,提出一种引入层次结构和三角形不等式原理的改进算法——IAHOC算法。将IAHOC算法分别应用于模拟数据和基准数据进行实验,比较分析得到的实验结果表明,IAHOC算法的层次结构能够更好地描述数据结构,并且辅以三角形不等式原理可以有效地降低原算法的计算复杂度。

In order to quickly and efficiently cluster the incremental data, in the paper we present an improved algorithm namely the IAHOC, which introduces the hierarchical structure and triangle inequality principle, based on AddC and AHOC algorithms to address the fiat structure deficiency of traditional online agglomerative clustering algorithm and the disadvantage in its execution speed. The IAHOC algorithm is applied to analog data and benchmark data separately in experiments, and the comparative analyses are made as well, the derived experimental results show that the hierarchical structure of IAHOC can describe data structure better, and with the help of the triangle inequality principle it can effectively reduce the computational complexity in original algorithm.