Mathematics > Statistics Theory
[Submitted on 19 Apr 2016 (this version), latest version 16 Apr 2019 (v5)]
Title:Optimal Bayesian posterior concentration rates with empirical priors
View PDFAbstract:In high-dimensional Bayesian applications, choosing a prior such that the corresponding posterior distribution has optimal asymptotic concentration properties can be restrictive in the sense that the priors supported by theory may not be computationally convenient, and vice versa. This paper develops a general strategy for constructing empirical or data-dependent priors whose corresponding posterior distributions have optimal concentration rate properties. The basic idea is to center the prior in a specific way on a good estimator. This makes the asymptotic properties of the posterior less sensitive to the shape of the prior which, in turn, allows users to work with priors of convenient forms while maintaining the optimal posterior concentration rates. General results on both adaptive and non-adaptive rates based on empirical priors are presented, along with illustrations in density estimation, nonparametric regression, and high-dimensional structured normal models.
Submission history
From: Ryan Martin [view email][v1] Tue, 19 Apr 2016 20:10:51 UTC (32 KB)
[v2] Wed, 25 May 2016 16:49:47 UTC (28 KB)
[v3] Mon, 1 May 2017 02:17:51 UTC (30 KB)
[v4] Wed, 6 Jun 2018 17:25:13 UTC (34 KB)
[v5] Tue, 16 Apr 2019 15:33:17 UTC (36 KB)
Current browse context:
math.ST
References & Citations
Bibliographic and Citation Tools
Bibliographic Explorer (What is the Explorer?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)
Code, Data and Media Associated with this Article
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)
Demos
Recommenders and Search Tools
Influence Flower (What are Influence Flowers?)
Connected Papers (What is Connected Papers?)
CORE Recommender (What is CORE?)
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.