A Bayesian nonparametric approach to log-concave density estimation
arXiv:1703.09531 · doi:10.3150/19-BEJ1139
Abstract
The estimation of a log-concave density on is a canonical problem in the area of shape-constrained nonparametric inference. We present a Bayesian nonparametric approach to this problem based on an exponentiated Dirichlet process mixture prior and show that the posterior distribution converges to the log-concave truth at the (near-) minimax rate in Hellinger distance. Our proof proceeds by establishing a general contraction result based on the log-concave maximum likelihood estimator that prevents the need for further metric entropy calculations. We also present two computationally more feasible approximations and a more practical empirical Bayes approach, which are illustrated numerically via simulations.
arXiv comment39 pages, 17 figures. Simulation studies were significantly expanded and one more theorem has been added
References in corpus (12)
- Rates of contraction of posterior distributions based on Gaussian process priors
- Maximum likelihood estimation of a log-concave density and its distribution function: Basic properties and uniform consistency
- Frequentist coverage of adaptive nonparametric Bayesian credible sets
- Posterior convergence rates of Dirichlet mixtures at smooth densities
- Limit distribution theory for maximum likelihood estimation of a log-concave density
- Approximation by log-concave distributions, with applications to regression
- Inference and Modeling with Log-concave Distributions
- Bayesian inverse problems with non-conjugate priors
- Nonparametric estimation of multivariate convex-transformed densities
- Bayesian nonparametric multivariate convex regression
- Concentration rate and consistency of the posterior under monotonicity constraints
- Smooth tail index estimation