Signal estimation problems with smoothness and sparsity priors can be naturally modeled as quadratic optimization with -"norm" constraints. Since such problems are non-convex and hard-to-solve, the standard approach is, instead, to tackle their convex surrogates based on -norm relaxations. In this paper, we propose a new iterative (convex) conic quadratic relaxations that exploit not only the -"norm" terms, but also the fitness and smoothness functions. The iterative convexification approach substantially closes the gap between the -"norm" and its surrogate. These stronger relaxations lead to significantly better estimators than -norm approaches and also allow one to utilize affine sparsity priors. In addition, the parameters of the model and the resulting estimators are easily interpretable. Experiments with a tailored Lagrangian decomposition method indicate that the proposed iterative convex relaxations \rev{yield solutions within 1\% of the exact approach, and can tackle instances with up to 100,000 variables under one minute.
View on arXiv