Title: Biclustering with Conjoined Dirichlet Processes
Abstract: Biclustering is a class of techniques that simultaneously clusters the rows and columns of a matrix to sort heterogeneous data into homogeneous blocks. Although many algorithms have been proposed to find biclusters, existing methods suffer from the pre-specification of the number of biclusters, or place constraints on the bicluster structure (e.g. no overlap, overlaps in only one direction). To address these issues, we are developing a non-parametric probabilistic biclustering method based on Dirichlet processes to identify biclusters with strong co-occurrence in both rows and columns. The proposed method utilizes two Dirichlet process mixture models to learn row and column clusters, with the number of resulting clusters determined by the data rather than pre-specified. Biclusters and any potential overlaps are identified by calculating the mutual dependence between the row and column clusters. Preliminary results demonstrate that our method improves bicluster extraction in several settings compared to existing approaches.