Convergence and Stability of a Class of Iteratively Re-weighted Least Squares Algorithms for Sparse Signal Recovery in the Presence of Noise

In this paper, we study the theoretical properties of a class of iteratively re-weighted least squares (IRLS) algorithms for sparse signal recovery in the presence of noise. We demonstrate a one-to-one correspondence between this class of algorithms and a class of Expectation-Maximization (EM) algor...

Celý popis

Uložené v:
Podrobná bibliografia
Vydané v:IEEE transactions on signal processing Ročník 62; číslo 1; s. 183 - 195
Hlavní autori: Babadi, Behtash, Ba, Demba, Purdon, Patrick L, Brown, Emery N
Médium: Journal Article
Jazyk:English
Vydavateľské údaje: United States 30.10.2013
ISSN:1053-587X
On-line prístup:Získať plný text
Tagy: Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
Popis
Shrnutí:In this paper, we study the theoretical properties of a class of iteratively re-weighted least squares (IRLS) algorithms for sparse signal recovery in the presence of noise. We demonstrate a one-to-one correspondence between this class of algorithms and a class of Expectation-Maximization (EM) algorithms for constrained maximum likelihood estimation under a Gaussian scale mixture (GSM) distribution. The IRLS algorithms we consider are parametrized by 0 < ≤ 1 and > 0. The EM formalism, as well as the connection to GSMs, allow us to establish that the IRLS( , ) algorithms minimize -smooth versions of the ℓ 'norms'. We leverage EM theory to show that, for each 0 < ≤ 1, the limit points of the sequence of IRLS( , ) iterates are stationary point of the -smooth ℓ 'norm' minimization problem on the constraint set. Finally, we employ techniques from Compressive sampling (CS) theory to show that the class of IRLS( , ) algorithms is stable for each 0 < ≤ 1, if the limit point of the iterates coincides the global minimizer. For the case = 1, we show that the algorithm converges exponentially fast to a neighborhood of the stationary point, and outline its generalization to super-exponential convergence for < 1. We demonstrate our claims via simulation experiments. The simplicity of IRLS, along with the theoretical guarantees provided in this contribution, make a compelling case for its adoption as a standard tool for sparse signal recovery.
Bibliografia:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 23
ISSN:1053-587X
DOI:10.1109/TSP.2013.2287685