9
58

Consistent Parameter Estimation for LASSO and Approximate Message Passing

Ali Mousavi
A. Maleki
Richard G. Baraniuk
Abstract

We consider the problem of recovering a vector βoRp\beta_o \in \mathbb{R}^p from nn random and noisy linear observations y=Xβo+wy= X\beta_o + w, where XX is the measurement matrix and ww is noise. The LASSO estimate is given by the solution to the optimization problem β^λ=argminβ12yXβ22+λβ1\hat{\beta}_{\lambda} = \arg \min_{\beta} \frac{1}{2} \|y-X\beta\|_2^2 + \lambda \| \beta \|_1. Among the iterative algorithms that have been proposed for solving this optimization problem, approximate message passing (AMP) has attracted attention for its fast convergence. Despite significant progress in the theoretical analysis of the estimates of LASSO and AMP, little is known about their behavior as a function of the regularization parameter λ\lambda, or the thereshold parameters τt\tau^t. For instance the following basic questions have not yet been studied in the literature: (i) How does the size of the active set β^λ0/p\|\hat{\beta}^\lambda\|_0/p behave as a function of λ\lambda? (ii) How does the mean square error β^λβo22/p\|\hat{\beta}_{\lambda} - \beta_o\|_2^2/p behave as a function of λ\lambda? (iii) How does βtβo22/p\|\beta^t - \beta_o \|_2^2/p behave as a function of τ1,,τt1\tau^1, \ldots, \tau^{t-1}? Answering these questions will help in addressing practical challenges regarding the optimal tuning of λ\lambda or τ1,τ2,\tau^1, \tau^2, \ldots. This paper answers these questions in the asymptotic setting and shows how these results can be employed in deriving simple and theoretically optimal approaches for tuning the parameters τ1,,τt\tau^1, \ldots, \tau^t for AMP or λ\lambda for LASSO. It also explores the connection between the optimal tuning of the parameters of AMP and the optimal tuning of LASSO.

View on arXiv
Comments on this paper