Semi-supervised agglomerative hierarchical clustering algorithms with pairwise constraints

摘要

Recently semi-supervised clustering has been studied by many researchers, but there are no extensive studies using different types of algorithms. In this paper we consider ag-glomerative hierarchical algorithms with pairwise constraints. The constraints are directly introduced to the single linkage which is equivalent to the transitive closure algorithm, while the centroid method and the Ward methods need kernelization of the algorithms. Simple numerical examples are shown to see how the constraints work.