Abstract
In this article, we consider the problem of clustering multi-view data, that is, information associated with individuals that form heterogeneous data sources (the views). We adopt a Bayesian model and in the prior structure we assume that each individual belongs to a baseline cluster, and conditionally allow each individual in each view to potentially belong to different clusters than the baseline. We call such a structure latent modularity.
Then, in each view, we have a specific statistical model with an associated prior. We derive expressions for the marginal priors on the view-specific cluster labels and the associated partitions, giving several insights into our chosen prior structure. Using a tailored Markov chain Monte Carlo algorithm, we showcase our model in a simulation study, as well as in a clinical case study involving multi-view measurements from the GUSTO cohort.
Joint work with M. De Iorio, G. Page, and A. Jasra