Distributed MCMC inference for Bayesian Non-Parametric Latent Block Model

Given a data matrix, where rows represent observations and columns represent variables or features, co-clustering, also known as bi-clustering aims to infer a row partition and a column partition simultaneously. The resulting partition is composed of homogeneous blocks. When a dataset exhibits a dual structure between observations and variables, co-clustering outperforms conventional clustering algorithms which only infers a row partition without considering the relationships between observations and variables. Co-clustering is a powerful data mining tool for two-dimensional data and is widely applied in various fields such as bioinformatics [1]. To tackle the co-clustering problem, the Latent Block Model (LBM) was introduced by [2].