Bayesian Model Comparison with the Hyvärinen Score: Computation and Consistency

DOI10.1080/01621459.2018.1518237zbMATH Open1428.62118arXiv1711.00136OpenAlexW3101515896MaRDI QIDQ5208087FDOQ5208087

Pierre E. Jacob, Stephane Shao, Jie Ding, Vahid Tarokh

Publication date: 15 January 2020

Published in: Journal of the American Statistical Association (Search for Journal in Brave)

Abstract: The Bayes factor is a widely used criterion in model comparison and its logarithm is a difference of out-of-sample predictive scores under the logarithmic scoring rule. However, when some of the candidate models involve vague priors on their parameters, the log-Bayes factor features an arbitrary additive constant that hinders its interpretation. As an alternative, we consider model comparison using the Hyv"arinen score. We propose a method to consistently estimate this score for parametric models, using sequential Monte Carlo methods. We show that this score can be estimated for models with tractable likelihoods as well as nonlinear non-Gaussian state-space models with intractable likelihoods. We prove the asymptotic consistency of this new model selection criterion under strong regularity assumptions in the case of non-nested models, and we provide qualitative insights for the nested case. We also use existing characterizations of proper scoring rules on discrete spaces to extend the Hyv"arinen score to discrete observations. Our numerical illustrations include L'evy-driven stochastic volatility models and diffusion models for population dynamics.

Full work available at URL: https://arxiv.org/abs/1711.00136

zbMATH Keywords

model selectionstate-space modelBayes factornoninformative priorSMC

Mathematics Subject Classification ID

Asymptotic properties of parametric estimators (62F12) Bayesian inference (62F15)

Cites Work

• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Non-Gaussian Ornstein-Uhlenbeck-based models and some of their uses in financial economics. (With discussion)
• Sequential Monte Carlo Samplers
• The Bayesian Choice
• Likelihood Ratio Tests for Model Selection and Non-Nested Hypotheses
• Time series analysis via mechanistic models
• Inference in hidden Markov models.
• Present Position and Potential Developments: Some Personal Views: Statistical Theory: The Prequential Approach
• Particle Markov Chain Monte Carlo Methods
• Bayes Factors
• SMC2: An Efficient Algorithm for Sequential Analysis of State Space Models
• Introduction to nonparametric estimation
• A sequential particle filter method for static models
• The Intrinsic Bayes Factor for Model Selection and Prediction
• A comment on D. V. Lindley's statistical paradox
• Consistency of objective Bayes factors as the model dimension grows
• Nonlinear time series. Theory, methods and applications with R examples
• Econometric Analysis of Realized Volatility and its Use in Estimating Stochastic Volatility Models
• Minimum scoring rule inference
• Efficient learning via simulation: a marginalized resample-move approach
• Dynamics of Bayesian updating with dependent data and misspecified models
• Bayesian model selection based on proper scoring rules
• Bayesian information in an experiment and the Fisher information distance
• MEASURES OF THE VALUE OF INFORMATION
• Proper local scoring rules
• Proper local scoring rules on discrete sample spaces
• Local proper scoring rules of order two
• Bounds on the deficit in the logarithmic Sobolev inequality
• Proper Scores for Probability Forecasters
• Consistency of Bayes estimators without the assumption that the model is correct
• About the posterior distribution in hidden Markov models with unknown number of states
• Bayes factors and marginal distributions in invariant situations
• On cross-validation of Bayesian models
• Bayesian inference with misspecified models
• An adaptive sequential Monte Carlo sampler
• Asymptotic behavior of Bayes estimators for hidden Markov models with application to ion channels
• An efficient computational approach for prior sensitivity analysis and cross‐validation
• Methods for Default and Robust Bayesian Model Comparison: The Fractional Bayes Factor Approach
• A note on Bayes factor consistency in partial linear models
• A note on Bayesian model selection for discrete data using proper scoring rules
• Bayesian consistency for stationary models
• OUP accepted manuscript

Cited In (9)

• Discussion of: ``Identifiability of latent-variable and structural-equation models: from linear to nonlinear
• An Invitation to Sequential Monte Carlo Samplers
• The Hyvärinen scoring rule in Gaussian linear time series models
• Comparison of stochastic frontier models using the Hyvärinen factor
• Robust inference methods for meta-analysis involving influential outlying studies
• Title not available (Why is that?)
• Title not available (Why is that?)
• Bayesian experimental design without posterior calculations: an adversarial approach
• A general approximation to nested Bayes factors with informed priors

This page was built for publication: Bayesian Model Comparison with the Hyvärinen Score: Computation and Consistency

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5208087)