Independence based clustering

Existing clustering methods focus on the similarity of data within the cluster. Therefore, distance and independence between clusters were not taken into account. However, users expect that the data within a cluster are similar, and data in different clusters are well separated or independent from each other. In this paper, we propose a clustering method where data within a cluster are similar, and data between clusters are highly independent. We show the results of experiments using benchmark data. And we carried out a survey with high school students.