ABC likelihood-free methods for model choice in Gibbs random fields
2009
Grelaud , Aude (INRA , Jouy-En-Josas (France). UR 1077 Mathématique, Informatique et Génome) | Robert , Christian P. (Université Paris Dauphine (Paris 9)Centre de Recherche en Economie et StatistiqueInstitut National des Statistiques et des Etudes Economiques, ParisParisParis(France).) | Marin , Jean-Michel (Université Montpellier 2Centre de Recherche en Economie et Statistique, Montpellier(France). Institut de Mathématiques et Modélisation de Montpellier) | Rodolphe , Francois (INRA , Jouy-En-Josas (France). UR 1077 Mathématique, Informatique et Génome) | Taly , Jean-Francois (INRA , Jouy-En-Josas (France). UR 1077 Mathématique, Informatique et Génome)
Gibbs random fields (GRF) are polymorphous statistical models that can be used to analyse di®erent types of dependence, in particular for spatially correlated data. However, when those models are faced with the challenge of selecting a dependence structure from many, the use of standard model choice methods is hampered by the unavailability of the normalising constant in the Gibbs likelihood. In particular, from a Bayesian perspective, the computation of the posterior probabilities of the models under competition requires special likelihood-free simulation techniques like the Approximate Bayesian Computation (ABC) algorithm that is intensively used in population genetics. We show in this paper how to implement an ABC algorithm geared towards model choice in the general setting of Gibbs random ¯elds, demonstrating in particular that there exists a su±cient statistic across models. The accuracy of the approximation to the posterior probabilities can be further improved by importance sampling on the distribution of the models. The practical aspects of the method are detailed through two applications, the test of an iid Bernoulli model versus a ¯rst-order Markov chain, and the choice of a folding structure for two proteins.
Show more [+] Less [-]AGROVOC Keywords
Bibliographic information
This bibliographic record has been provided by Institut national de la recherche agronomique
