Calibrated Generalized Bayesian Inference

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Frazier, David T., Drovandi, Christopher, Kohn, Robert
Formato: Preprint
Publicado: 2023
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866908874531930112
author Frazier, David T.
Drovandi, Christopher
Kohn, Robert
author_facet Frazier, David T.
Drovandi, Christopher
Kohn, Robert
contents We propose a simple approach that provides accurate uncertainty quantification for Bayesian inference in misspecified or approximate models, and for generalized (Gibbs) posteriors. While existing solutions in this context are based on explicit Gaussian approximations or post-processing procedures, we demonstrate that correct uncertainty quantification can be achieved by substituting the usual posterior with an intuitively appealing alternative that conveys the same information. This solution applies to both likelihood-based and loss-based posteriors, and is formally demonstrated to reliably quantify uncertainty. This new approach is demonstrated through a range of examples, including generalized linear models, and doubly intractable models.
format Preprint
id arxiv_https___arxiv_org_abs_2311_15485
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Calibrated Generalized Bayesian Inference
Frazier, David T.
Drovandi, Christopher
Kohn, Robert
Methodology
We propose a simple approach that provides accurate uncertainty quantification for Bayesian inference in misspecified or approximate models, and for generalized (Gibbs) posteriors. While existing solutions in this context are based on explicit Gaussian approximations or post-processing procedures, we demonstrate that correct uncertainty quantification can be achieved by substituting the usual posterior with an intuitively appealing alternative that conveys the same information. This solution applies to both likelihood-based and loss-based posteriors, and is formally demonstrated to reliably quantify uncertainty. This new approach is demonstrated through a range of examples, including generalized linear models, and doubly intractable models.
title Calibrated Generalized Bayesian Inference
topic Methodology
url https://arxiv.org/abs/2311.15485