Bayesian Analysis of Simple Random Densities

TL;DR

This paper introduces a new nonparametric Bayesian prior for densities that is computationally manageable, maintains its form under sampling, and has well-behaved posterior properties.

Contribution

It proposes a novel nonparametric prior for densities that is both tractable and closed under sampling, with proven posterior asymptotics.

Findings

Prior is computationally feasible

Model is closed under sampling

Posterior exhibits proper asymptotic behavior

Abstract

A tractable nonparametric prior over densities is introduced which is closed under sampling and exhibits proper posterior asymptotics.