Robust Bayesian synthetic likelihood via a semi-parametric approach

DOI10.48550/ARXIV.1809.05800zbMATH Open1436.62090arXiv1809.05800OpenAlexW2978753741MaRDI QIDQ91059FDOQ91059

C. C. Drovandi, Ziwen An, David J. Nott, David J. Nott, Ziwen An

Publication date: 16 September 2018

Published in: Statistics and Computing (Search for Journal in Brave)

Abstract: Bayesian synthetic likelihood (BSL) is now a well established method for performing approximate Bayesian parameter estimation for simulation-based models that do not possess a tractable likelihood function. BSL approximates an intractable likelihood function of a carefully chosen summary statistic at a parameter value with a multivariate normal distribution. The mean and covariance matrix of this normal distribution are estimated from independent simulations of the model. Due to the parametric assumption implicit in BSL, it can be preferred to its non-parametric competitor, approximate Bayesian computation, in certain applications where a high-dimensional summary statistic is of interest. However, despite several successful applications of BSL, its widespread use in scientific fields may be hindered by the strong normality assumption. In this paper, we develop a semi-parametric approach to relax this assumption to an extent and maintain the computational advantages of BSL without any additional tuning. We test our new method, semiBSL, on several challenging examples involving simulated and real data and demonstrate that semiBSL can be significantly more robust than BSL and another approach in the literature.

Full work available at URL: https://arxiv.org/abs/1809.05800

Recommendations

• Variational Bayes with synthetic likelihood
• Bayesian parametric bootstrap for models with intractable likelihoods
• Constructing Summary Statistics for Approximate Bayesian Computation: Semi-Automatic Approximate Bayesian Computation
• Likelihood-free inference in high dimensions with synthetic likelihood
• An extended empirical saddlepoint approximation for intractable likelihoods

zbMATH Keywords

copulaapproximate Bayesian computation (ABC)kernel density estimationlikelihood-free inferencenonparanormal distributionrobust estimation

Mathematics Subject Classification ID

Bayesian inference (62F15) Characterization and structure theory for multivariate probability distributions; copulas (62H05) Robustness and adaptive procedures (parametric inference) (62F35)

Cites Work

• An extended empirical saddlepoint approximation for intractable likelihoods
• Bayesian Synthetic Likelihood
• Accelerating Bayesian Synthetic Likelihood With the Graphical Lasso
• A Method for Simulating Stable Random Variables
• The nonparanormal: semiparametric estimation of high dimensional undirected graphs
• Title not available (Why is that?)
• Variable kernel density estimation
• The pseudo-marginal approach for efficient Monte Carlo computations
• The largest inclusions in a piece of steel
• Sparse inverse covariance estimation with the graphical lasso
• Extending approximate Bayesian computation methods to high dimensions via a Gaussian copula model
• Likelihood-free inference via classification
• Variational Bayes with synthetic likelihood
• Likelihood-free inference in high dimensions with synthetic likelihood
• Vines -- a new graphical model for dependent random variables.
• Bayesian indirect inference using a parametric auxiliary model
• Approximate Bayesian computation by subset simulation
• Remarks on Some Nonparametric Estimates of a Density Function
• Penalized Normal Likelihood and Ridge Regularization of Correlation and Covariance Matrices
• Sinh-arcsinh distributions
• A new class of multivariate skew distributions with applications to bayesian regression models
• Constructing Summary Statistics for Approximate Bayesian Computation: Semi-Automatic Approximate Bayesian Computation
• Inference for Stereological Extremes
• On Estimation of a Probability Density Function and Mode
• The Gaussian rank correlation estimator: robustness properties

Cited In (15)

• Multifidelity multilevel Monte Carlo to accelerate approximate Bayesian parameter inference for partially observed stochastic processes
• Transformations in semi-parametric Bayesian synthetic likelihood
• Parallel Gaussian process surrogate Bayesian inference with noisy likelihood evaluations
• Generalized Bayesian likelihood-free inference
• BSL
• Bayesian Synthetic Likelihood
• Robust Approximate Bayesian Inference With Synthetic Likelihood
• A comparison of likelihood-free methods with and without summary statistics
• Robust Bayesian Credibility Using Semiparametric Models
• Title not available (Why is that?)
• Summary statistics and discrepancy measures for approximate Bayesian computation via surrogate posteriors
• Bayesian Inference Using Synthetic Likelihood: Asymptotics and Adjustments
• Finding our way in the dark: approximate MCMC for approximate Bayesian methods
• Sequentially guided MCMC proposals for synthetic likelihoods and correlated synthetic likelihoods
• Efficient Bayesian Synthetic Likelihood With Whitening Transformations

Uses Software

• glasso
• ABC-SubSim

This page was built for publication: Robust Bayesian synthetic likelihood via a semi-parametric approach

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q91059)