Approximating predictive probabilities of Gibbs-type priors - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2017

Approximating predictive probabilities of Gibbs-type priors

Résumé

Gibbs-type priors are arguably the most 'natural' generalization of the Dirichlet prior. Among them the two parameter Poisson-Dirichlet prior certainly stands out for the simplicity and intuitiveness of its predictive probabilities. Given an observable sample of size $n$, in this paper we show that the predictive probabilities of any Gibbs-type prior admit a large $n$ approximation, with an error term vanishing as $o(1/n)$, which maintains the same mathematical tractability and interpretability as the predictive probabilities of the two parameter Poisson-Dirichlet prior. We discuss the use of our approximate predictive probabilities in connection with some recent work on Bayesian nonparametric inference for discovery probabilities.
Fichier non déposé

Dates et versions

hal-01667746 , version 1 (19-12-2017)

Identifiants

  • HAL Id : hal-01667746 , version 1

Citer

Julyan Arbel. Approximating predictive probabilities of Gibbs-type priors. ERCIM - 10th International Conference of the ERCIM WG on Computational and Methodological Statistics, Dec 2017, London, United Kingdom. ⟨hal-01667746⟩
90 Consultations
0 Téléchargements

Partager

Gmail Facebook X LinkedIn More