Objective Bayesian inference with proper scoring rules

DOI10.1007/S11749-018-0597-ZzbMATH Open1420.62112arXiv1711.10819OpenAlexW2772314143MaRDI QIDQ2273176FDOQ2273176

Authors: Valentina Mameli, Federica Giummolè, Erlis Ruli, Laura Ventura 

Publication date: 18 September 2019

Published in: Test (Search for Journal in Brave)

Abstract: Standard Bayesian analyses can be difficult to perform when the full likelihood, and consequently the full posterior distribution, is too complex and difficult to specify or if robustness with respect to data or to model misspecifications is required. In these situations, we suggest to resort to a posterior distribution for the parameter of interest based on proper scoring rules. Scoring rules are loss functions designed to measure the quality of a probability distribution for a random variable, given its observed value. Important examples are the Tsallis score and the Hyv"arinen score, which allow us to deal with model misspecifications or with complex models. Also the full and the composite likelihoods are both special instances of scoring rules. The aim of this paper is twofold. Firstly, we discuss the use of scoring rules in the Bayes formula in order to compute a posterior distribution, named SR-posterior distribution, and we derive its asymptotic normality. Secondly, we propose a procedure for building default priors for the unknown parameter of interest that can be used to update the information provided by the scoring rule in the SR-posterior distribution. In particular, a reference prior is obtained by maximizing the average $alpha-$divergence from the SR-posterior distribution. For $0leq|alpha|<1$, the result is a Jeffreys-type prior that is proportional to the square root of the determinant of the Godambe information matrix associated to the scoring rule. Some examples are discussed.

Full work available at URL: https://arxiv.org/abs/1711.10819

Recommendations

• On a class of objective priors from scoring rules (with discussion)
• Preposterior expected loss as a scoring rule for prior distributions
• Proper local scoring rules on discrete sample spaces
• Theory and applications of proper scoring rules
• Bayesian model selection based on proper scoring rules

zbMATH Keywords

composite likelihoodreference priorrobustnessscoring rule\(\alpha\)-divergencesgodambe information\(M\)-estimating function

Mathematics Subject Classification ID

Bayesian inference (62F15) Asymptotic distribution theory in statistics (62E20) Robustness and adaptive procedures (parametric inference) (62F35)

Cites Work

• Title not available (Why is that?)
• Bayesian inference from composite likelihoods, with an application to spatial extremes
• Title not available (Why is that?)
• An Optimum Property of Regular Maximum Likelihood Estimation
• Pseudo-Likelihoods for Bayesian Inference
• Bayesian composite marginal likelihoods
• Title not available (Why is that?)
• Title not available (Why is that?)
• Robust Statistics
• Inference for clustered data using the independence loglikelihood
• An extended Gaussian max-stable process model for spatial extremes
• Title not available (Why is that?)
• Bayesian empirical likelihood for quantile regression
• Some extensions of score matching
• Estimation of non-normalized statistical models by score matching
• Robust likelihood functions in Bayesian inference
• Minimum scoring rule inference
• Approximate Bayesian computation with composite score functions
• Bayesian empirical likelihood
• Default prior distributions from quasi- and quasi-profile likelihoods
• The formal definition of reference priors
• Possible generalization of Boltzmann-Gibbs statistics.
• A parametric framework for the comparison of methods of very robust regression
• Rejoinder: ``The case for objective Bayesian analysis
• Robust Bayes estimation using the density power divergence
• Title not available (Why is that?)
• Robust and efficient estimation by minimising a density power divergence
• Theory and applications of proper scoring rules
• Overall objective priors
• Bayesian model selection based on proper scoring rules
• Bayesian information in an experiment and the Fisher information distance
• Objective priors: an introduction for frequentists
• Proper local scoring rules
• Quasi Bayesian likelihood
• A general divergence criterion for prior selection
• Bayesian exponentially tilted empirical likelihood
• A characterization of monotone and regular divergences
• Adjusting composite likelihood ratio statistics
• Higher-order asymptotics for scoring rules
• Probability matching priors: Higher order asymptotics
• Contrasting probabilistic scoring rules
• The geometry of proper scoring rules
• Bayesian and frequentist confidence intervals arising from empirical-type likelihoods
• Title not available (Why is that?)
• Characterization of priors under which Bayesian and frequentist Bartlett corrections are equivalent in the multiparameter case
• On divergence measures leading to Jeffreys and other reference priors
• Bootstrap adjustments of signed scoring rule root statistics

Cited In (10)

• Bayesian model selection based on proper scoring rules
• Objective Bayes, conditional inference and the signed root likelihood ratio statistic
• Recent advances in directional statistics
• Minimum scoring rule inference
• Generalized Bayesian Inference for Discrete Intractable Likelihood
• Generalized Bayesian likelihood-free inference
• On a class of objective priors from scoring rules (with discussion)
• Preposterior expected loss as a scoring rule for prior distributions
• Robust inference for non-linear regression models from the Tsallis score: application to coronavirus disease 2019 Contagion in Italy
• Robust confidence distributions from proper scoring rules

This page was built for publication: Objective Bayesian inference with proper scoring rules

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2273176)