Precise performance analysis of the LASSO under matrix uncertainties

Abstract
In this paper, we consider the problem of recovering an unknown sparse signal x ∈ R from noisy linear measurements {equation presented}. A popular approach is to solve the ℓ-norm regularized least squares problem which is known as the LASSO. In many practical situations, the measurement matrix H is not perfectely known and we only have a noisy version of it. We assume that the entries of the measurement matrix H and of the noise vector z are iid Gaussian with zero mean and variances 1 /n and σ . In this work, an imperfect measurement matrix is considered under which we precisely characterize the limilting behavior of the mean squared error and the probability of support recovery of the LASSO. The analysis is performed when the problem dimensions grow simultaneously to infinity at fixed rates. Numerical simulations validate the theoretical predictions derived in this paper.

Citation
Alrashdi AM, Ben Atitallah I, Al-Naffouri TY, Alouini M-S (2017) Precise performance analysis of the LASSO under matrix uncertainties. 2017 IEEE Global Conference on Signal and Information Processing (GlobalSIP). Available: http://dx.doi.org/10.1109/GlobalSIP.2017.8309169.

Publisher
Institute of Electrical and Electronics Engineers (IEEE)

Journal
2017 IEEE Global Conference on Signal and Information Processing (GlobalSIP)

Conference/Event Name
5th IEEE Global Conference on Signal and Information Processing, GlobalSIP 2017

DOI
10.1109/GlobalSIP.2017.8309169

arXiv
1808.04309

Additional Links
https://ieeexplore.ieee.org/document/8309169/

Permanent link to this record