Abstract:We study the minimization problem of a non-convex sparsity promoting penalty function, the transformed $l_1$ (TL1), and its application in compressed sensing (CS). The TL1 penalty interpolates $l_0$ and $l_1$ norms through a nonnegative parameter $a \in (0,+\infty)$, similar to $l_p$ with $p \in (0,1]$, and is known to satisfy unbiasedness, sparsity and Lipschitz continuity properties. We first consider the constrained minimization problem and discuss the exact recovery of $l_0$ norm minimal solution based on the null space property (NSP). We then prove the stable recovery of $l_0$ norm minimal solution if the sensing matrix $A$ satisfies a restricted isometry property (RIP). Next, we present difference of convex algorithms for TL1 (DCATL1) in computing TL1-regularized constrained and unconstrained problems in CS. The inner loop concerns an $l_1$ minimization problem on which we employ the Alternating Direction Method of Multipliers (ADMM). For the unconstrained problem, we prove convergence of DCATL1 to a stationary point satisfying the first order optimality condition. In numerical experiments, we identify the optimal value $a=1$, and compare DCATL1 with other CS algorithms on two classes of sensing matrices: Gaussian random matrices and over-sampled discrete cosine transform matrices (DCT). We find that for both classes of sensing matrices, the performance of DCATL1 algorithm (initiated with $l_1$ minimization) always ranks near the top (if not the top), and is the most robust choice insensitive to the conditioning of the sensing matrix $A$. DCATL1 is also competitive in comparison with DCA on other non-convex penalty functions commonly used in statistics with two hyperparameters.

Smoothed sparse recovery via locally competitive algorithm and forward Euler discretization method.

On Partial Smoothness, Activity Identification and Faster Algorithms of L1 over L2 Minimization

On Partly Smoothness, Activity Identification and Faster Algorithms of L 1 over L 2 Minimization

On the Convergence of an Active-Set Method for ℓ1minimization

ADMM for Nonconvex Optimization under Minimal Continuity Assumption

A Riemannian ADMM

An Alternating Direction Method with Continuation for Nonconvex Low Rank Minimization

A fast ADMM algorithm for sparse precision matrix estimation using lasso penalized D-trace loss

Nonlinear Eigen-approach ADMM for Sparse Optimization on Stiefel Manifold

Study on L_1 over L_2 Minimization for Nonnegative Signal Recovery

Study on <i>L</i><sub>1</sub> over <i>L</i><sub>2</sub> Minimization for Nonnegative Signal Recovery

An over-relaxed ADMM for separable convex programming and its applications to statistical learning

A Unified Study on $L_1$ over $L_2$ Minimization

Adaptive finite element approximation of sparse optimal control with integral fractional Laplacian

Minimization of Transformed $L_1$ Penalty: Theory, Difference of Convex Function Algorithm, and Robust Application in Compressed Sensing

Unified Analysis on L1 over L2 Minimization for Signal Recovery

On the implementation of ADMM with dynamically configurable parameter for the separable minimization

Image Smoothing Via Gradient Sparsity and Surface Area Minimization

A partial Bregman ADMM with a general relaxation factor for structured nonconvex and nonsmooth optimization

A Unified Inexact Stochastic ADMM for Composite Nonconvex and Nonsmooth Optimization

Sequential inertial linear ADMM algorithm for nonconvex and nonsmooth multiblock problems with nonseparable structure