Abstract:Multivariate, heteroscedastic errors complicate statistical inference in many large-scale denoising problems. Empirical Bayes is attractive in such settings, but standard parametric approaches rest on assumptions about the form of the prior distribution which can be hard to justify and which introduce unnecessary tuning parameters. We extend the nonparametric maximum likelihood estimator (NPMLE) for Gaussian location mixture densities to allow for multivariate, heteroscedastic errors. NPMLEs estimate an arbitrary prior by solving an infinite-dimensional, convex optimization problem; we show that this convex optimization problem can be tractably approximated by a finite-dimensional version. The empirical Bayes posterior means based on an NPMLE have low regret, meaning they closely target the oracle posterior means one would compute with the true prior in hand. We prove an oracle inequality implying that the empirical Bayes estimator performs at nearly the optimal level (up to logarithmic factors) for denoising without prior knowledge. We provide finite-sample bounds on the average Hellinger accuracy of an NPMLE for estimating the marginal densities of the observations. We also demonstrate the adaptive and nearly-optimal properties of NPMLEs for deconvolution. We apply our method to two denoising problems in astronomy, constructing a fully data-driven color-magnitude diagram of 1.4 million stars in the Milky Way and investigating the distribution of 19 chemical abundance ratios for 27 thousand stars in the red clump. We also apply our method to hierarchical linear models, illustrating the advantages of nonparametric shrinkage of regression coefficients on an education data set and on a microarray data set.

A Bayesian estimator for correlation parameters of the multivariate nakagami-m distribution

A Simulation-Based Approach to Estimate the Correlation Parameters of Multivariate Nakagami-m Distribution.

Generalized Selection Combining in Generalized Correlated Nakagami-m Fading

Performance Analysis of Generalized Selection Combining in Generalized Correlated Nakagami-m Fading.

Novel ML Estimation of M Parameter of the Noisy Nakagami-m Channel

Analysis of multivariate Gaussian mixture model based on DRJMCMC method

Bayesian estimation for longitudinal data in a joint model with HPCs

Complex-valued Bayesian parameter estimation via Markov chain Monte Carlo

Statistical analysis of cascaded Nakagami-m fading channels with generalized correlation

Bayesian mixtures of common factor analyzers: Model, variational inference, and applications

Multivariate, Heteroscedastic Empirical Bayes via Nonparametric Maximum Likelihood

Bayesian mixture modeling for multivariate conditional distributions

Bayesian analysis of mixtures of lognormal distribution with an unknown number of components from grouped data

POSTERIOR ANALYSIS OF THE MULTIPLICATIVE HETEROSCEDASTICITY MODEL

Bayesian estimation of correlation functions

A Bayesian approach for clustering skewed data using mixtures of multivariate normal-inverse Gaussian distributions

Incorporating Grouping Information into Bayesian Gaussian Graphical Model Selection

On Efficient and Scalable Computation of the Nonparametric Maximum Likelihood Estimator in Mixture Models

Distributed Bayesian Posterior Sampling Via Moment Sharing

Parallel Markov Chain Monte Carlo for Non-Gaussian Posterior Distributions

Bayesian Nonparametric Covariance Regression