Information Theoretic Hierarchical Clustering

摘要: Hierarchical clustering has been extensively used in practice, where clusters can be assigned and analyzed simultaneously, especially when estimating the number of is challenging. However, due to conventional proximity measures recruited these algorithms, they are only capable detecting mass-shape encounter problems identifying complex data structures. Here, we introduce two bottom-up hierarchical approaches that exploit an information theoretic measure explore nonlinear boundaries between extract structures further than second order statistics. Experimental results on both artificial real datasets demonstrate superiority proposed algorithm compared algorithms reported literature, true clusters.