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

Download:

Current browse context:

stat

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: Generalized SURE for optimal shrinkage of singular values in low-rank matrix denoising

Abstract: We consider the problem of estimating a low-rank signal matrix from noisy measurements under the assumption that the distribution of the data matrix belongs to an exponential family. In this setting, we derive generalized Stein's unbiased risk estimation (SURE) formulas that hold for any spectral estimators which shrink or threshold the singular values of the data matrix. This leads to new data-driven spectral estimators, whose optimality is discussed using tools from random matrix theory and through numerical experiments. Under the spiked population model and in the asymptotic setting where the dimensions of the data matrix are let going to infinity, some theoretical properties of our approach are compared to recent results on asymptotically optimal shrinking rules for Gaussian noise. It also leads to new procedures for singular values shrinkage in finite-dimensional matrix denoising for Gamma-distributed and Poisson-distributed measurements.
Subjects: Statistics Theory (math.ST)
MSC classes: 62H12, 62H25
Cite as: arXiv:1605.07412 [math.ST]
  (or arXiv:1605.07412v3 [math.ST] for this version)

Submission history

From: Jeremie Bigot [view email]
[v1] Tue, 24 May 2016 12:33:45 GMT (897kb,D)
[v2] Sat, 22 Apr 2017 20:02:59 GMT (2797kb,D)
[v3] Mon, 2 Oct 2017 11:08:57 GMT (3451kb,D)

Link back to: arXiv, form interface, contact.