Composite likelihood inference in a discrete latent variable model for\n two-way "clustering-by-segmentation" problems

2015/06/27 by Francesco Bartolucci, Francesca Chiaromonte, Bartolucci, Francesco +5 · 1 citation
Biochemistry, Genetics and Molecular Biology · Computer Science · Mathematics · #Bayesian Methods and Mixture Models #FOS: Computer and information sciences #FOS: Mathematics #Gene expression and cancer classification #Methodology (stat.ME) #Statistical Methods and Inference #Statistics Theory (math.ST)

paper · pdf · doi:10.48550/arxiv.1506.08278

openalex publication_date 2015/06/27 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

We consider a discrete latent variable model for two-way data arrays, which\nallows one to simultaneously produce clusters along one of the data dimensions\n(e.g. exchangeable observational units or features) and contiguous groups, or\nsegments, along the other (e.g. consecutively ordered times or locations). The\nmodel relies on a hidden Markov structure but, given its complexity, cannot be\nestimated by full maximum likelihood. We therefore introduce composite\nlikelihood methodology based on considering different subsets of the data. The\nproposed approach is illustrated by simulation, and with an application to\ngenomic data.\n

Cited by

• Dealing with separation problem in hidden Markov models with covariates based on a penalized maximum likelihood approach

Related

• Composite empirical likelihood for multisample clustered data
• Computing the likelihood of sequence segmentation under Markov modelling
• Clustering With Side Information: From a Probabilistic Model to a Deterministic Algorithm
• Optimal Bayesian estimators for latent variable cluster models
• A Bayesian Approach to Restricted Latent Class Models for Scientifically-Structured Clustering of Multivariate Binary Outcomes