We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

math.ST

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Mathematics > Statistics Theory

Title: Data-driven priors and their posterior concentration rates

Abstract: In high-dimensional problems, choosing a prior distribution such that the corresponding posterior has desirable practical and theoretical properties can be challenging. This begs the question: can the data be used to help choose a good prior? In this paper, we develop a general strategy for constructing a data-driven or empirical prior and sufficient conditions for the corresponding posterior distribution to achieve a certain concentration rate. The idea is that the prior should put sufficient mass on parameter values for which the likelihood is large. An interesting byproduct of this data-driven centering is that the asymptotic properties of the posterior are less sensitive to the prior shape which, in turn, allows users to work with priors of computationally convenient forms while maintaining the desired 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.
Comments: 31 pages
Subjects: Statistics Theory (math.ST)
Journal reference: Electronic Journal of Statistics, 2019, volume 13, number 2, pages 3049--3081
DOI: 10.1214/19-EJS1600
Cite as: arXiv:1604.05734 [math.ST]
  (or arXiv:1604.05734v5 [math.ST] for this version)

Submission history

From: Ryan Martin [view email]
[v1] Tue, 19 Apr 2016 20:10:51 GMT (32kb)
[v2] Wed, 25 May 2016 16:49:47 GMT (28kb)
[v3] Mon, 1 May 2017 02:17:51 GMT (30kb)
[v4] Wed, 6 Jun 2018 17:25:13 GMT (34kb)
[v5] Tue, 16 Apr 2019 15:33:17 GMT (36kb)

Link back to: arXiv, form interface, contact.