Nonparametric Bayesian analysis of the compound Poisson prior for support boundary recovery

DOI10.1214/19-AOS1853MaRDI QIDQ2196227zbMATH OpenOpenAlexFDO

Authors Markus Reiß, Johannes Schmidt-Hieber

Publication date 28 August 2020

Published in The Annals of Statistics (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1809.04140, https://projecteuclid.org/euclid.aos/1594972824

zbMATH Keywords

Bernstein-von Mises theoremboundary detectioncompound Poisson processfrequentist Bayes analysisPoisson point processposterior contractionsubordinator prior

Mathematics Subject Classification ID

Point processes (e.g., Poisson, Cox, Hawkes processes) (60G55) Bayesian problems; characterization of Bayes procedures (62C10) Nonparametric estimation (62G05)

Abstract: Given data from a Poisson point process with intensity $(x,y)mapstonmathbf1(f(x)leqy),$ frequentist properties for the Bayesian reconstruction of the support boundary function $f$ are derived. We mainly study compound Poisson process priors with fixed intensity proving that the posterior contracts with nearly optimal rate for monotone and piecewise constant support boundaries and adapts to H"older smooth boundaries with smoothness index at most one. We then derive a non-standard Bernstein-von Mises result for a compound Poisson process prior and a function space with increasing parameter dimension. As an intermediate result the limiting shape of the posterior for random histogram type priors is obtained. In both settings, it is shown that the marginal posterior of the functional $vartheta=intf$ performs an automatic bias correction and contracts with a faster rate than the MLE. In this case, $(1-alpha)$-credible sets are also asymptotic $(1-alpha)$-confidence intervals. As a negative result, it is shown that the frequentist coverage of credible sets is lost for linear functions indicating that credible sets only have frequentist coverage for priors that are specifically constructed to match properties of the underlying true function.

Review text

The paper presents the performance of compound Poisson (CPP) as nonparametric priors. It is proved that under (CPP) priors, optimal posterior contraction rates are attained for Hölder functions and for monotone functions. In Section 2, the contraction rates for compound Poisson processes and subordinator priors are investigated. A general description of the asymptotic posterior shape in which the results thereafter can be embedded is presented in Section 3. In Section 4, Bernstein-von Mises-type theorems and results on the frequentist coverage of credible sets for CPP priors are presented. The proofs of the new theorems are presented in two appendices.

Recommendations

• Nonparametric Bayesian inference for multidimensional compound Poisson processes
• Compound Poisson processes, latent shrinkage priors and Bayesian nonconvex penalization
• An algorithm for nonparametric Bayesian estimation of a Poisson intensity
• Non-parametric Bayesian Estimation of a Spatial Poisson Intensity
• Bayesian inferences on nonlinear functions of Poisson rates
• Posterior contraction rates for support boundary recovery
• Bayesian computation for the superposition of nonhomogeneous poisson processes

Cites work

• A Bayesian CART algorithm
• A Bayesian nonparametric approach to log-concave density estimation
• A Bernstein-von Mises theorem for smooth functionals in semiparametric models
• A Bernstein-von Mises theorem in the nonparametric right-censoring model
• Adaptive function estimation in nonparametric regression with one-sided errors
• Asymptotic equivalence for nonparametric regression with non-regular errors
• Asymptotic normality of posterior distributions for exponential families when the number of parameters tends to infinity.
• Asymptotic normality of posterior distributions in high-dimensional linear models
• Asymptotic Statistics
• BART: Bayesian additive regression trees
• Bayesian detection of image boundaries
• Bayesian linear regression with sparse priors
• Bernstein-von Mises theorem for linear functionals of the density
• Bernstein-von Mises theorems for Gaussian regression with increasing number of regressors
• Concentration rate and consistency of the posterior distribution for selected priors under monotonicity constraints
• Consistency of Bayes estimators of a binary regression function
• Efficient estimation of functionals in nonparametric boundary models
• Finite sample Bernstein-von Mises theorem for semiparametric problems
• Fundamentals of nonparametric Bayesian inference
• scientific article; zbMATH DE number 1026574 (Why is no real title available?)
• Likelihood Estimation and Inference in a Class of Nonregular Econometric Models
• Minimax theory of image reconstruction
• Multiscale change point inference. With discussion and authors' reply
• Needles and straw in a haystack: posterior concentration for possibly sparse sequences
• On convergence of posterior distributions
• On Estimation of Monotone and Concave Frontier Functions
• On Rates of Convergence for Bayesian Density Estimation
• On risk bounds in isotonic and other shape restricted regression problems
• Small ball estimates in \(p\)-variation for stable processes
• Weak convergence and empirical processes. With applications to statistics

Cited in

(10)

• Criteria for posterior consistency and convergence at a rate
• Oracle posterior contraction rates under hierarchical priors
• Posterior contraction rates for support boundary recovery
• Bayesian variance estimation in the Gaussian sequence model with partial information on the means
• A Bayesian nonparametric approach to log-concave density estimation
• Functional estimation and hypothesis testing in nonparametric boundary models
• Bernstein-von Mises theorems for statistical inverse problems. II: Compound Poisson processes
• Compound Poisson processes, latent shrinkage priors and Bayesian nonconvex penalization
• Bayesian boundary trend filtering
• Nonparametric Bayesian estimation in a multidimensional diffusion model with high frequency data

Describes a project that uses

Uses Software

• BartPy

This page was built for publication: Nonparametric Bayesian analysis of the compound Poisson prior for support boundary recovery

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2196227)