Enhanced Sparsity by Non-Separable Regularization
arXiv:1511.00721 · doi:10.1109/TSP.2016.2518989
Abstract
This paper develops a convex approach for sparse one-dimensional deconvolution that improves upon L1-norm regularization, the standard convex approach. We propose a sparsity-inducing non-separable non-convex bivariate penalty function for this purpose. It is designed to enable the convex formulation of ill-conditioned linear inverse problems with quadratic data fidelity terms. The new penalty overcomes limitations of separable regularization. We show how the penalty parameters should be set to ensure that the objective function is convex, and provide an explicit condition to verify the optimality of a prospective solution. We present an algorithm (an instance of forward-backward splitting) for sparse deconvolution using the new penalty.
arXiv comment38 pages
References in corpus (12)
- Fast Image Recovery Using Variable Splitting and Constrained Optimization
- A fast approach for overcomplete sparse decomposition based on smoothed L0 norm
- One-step sparse estimates in nonconcave penalized likelihood models
- Sparse Signal Reconstruction via Iterative Support Detection
- Group-Sparse Signal Denoising: Non-Convex Regularization, Convex Optimization
- Sparse Signal Estimation by Maximally Sparse Convex Optimization
- Euclid in a Taxicab: Sparse Blind Deconvolution with Smoothed l1/l2 Regularization
- A Majorize-Minimize subspace approach for l2-l0 image regularization
- On the Convergence of the Iterative Shrinkage/Thresholding Algorithm With a Weakly Convex Penalty
- Convex Denoising using Non-Convex Tight Frame Regularization
- Homotopy based algorithms for -regularized least-squares
- Iterative Log Thresholding
Cited by in corpus (6)
- Sparse Regularization via Convex Analysis
- Total Variation Denoising via the Moreau Envelope
- A Penalty Function Promoting Sparsity Within and Across Groups
- Linearly-involved Moreau-Enhanced-over-Subspace Model: Debiased Sparse Modeling and Stable Outlier-Robust Regression
- Basis Pursuit Denoise with Nonsmooth Constraints
- Monotone Lipschitz-Gradient Denoiser: Explainability of Operator Regularization Approaches Free From Lipschitz Constant Control