We propose a Bayesian dual clustering method that infers both data and parameter granularities.
Kim et al. (Thu,) studied this question.