Applying Cluster Ensemble to Adaptive Tree Structured Clustering

Yamaguchi Takashi
Noguchi Yuki
Ichimura Takumi
Mackin Kenneth J.
Adaptive tree structured clustering (ATSC) is our proposed divisive hierarchical clustering method that recursively divides a data set into 2 subsets using self-organizing feature map (SOM). In each partition, the data set is quantized by SOM and the quantized data is divided using agglomerative hierarchical clustering. ATSC can divide data sets regardless of data size in feasible time. On the other hand clustering result stability of ATSC is equally unstable as other divisive hierarchical clustering and partitioned clustering methods. In this paper, we apply cluster ensemble for each data partition of ATSC in order to improve stability. Cluster ensemble is a framework for improving partitioned clustering stability. As a result of applying cluster ensemble, ATSC yields unique clustering results that could not be yielded by previous hierarchical clustering methods. This is because a different class distances function is used in each division in ATSC.