An iterative thresholding algorithm for linear inverse problems with a sparsity constraint
Ingrid C. Daubechies, Michel Defrise, Christine De Mol · Communications on Pure and Applied Mathematics · 2004
Abstract We consider linear inverse problems where the solution is assumed to have a sparse expansion on an arbitrary preassigned orthonormal basis. We prove that replacing the usual quadratic regularizing penalties by weighted 𝓁 p ‐penalties on the coefficients of such expansions, with 1 ≤ p ≤ 2, still regularizes the problem. Use of such 𝓁 p ‐penalized problems with p < 2 is often advocated when one expects the underlying ideal noiseless solution to have a sparse expansion with respect to the basis under consideration. To compute the corresponding regularized solutions, we analyze an iterative algorithm that amounts to a Landweber iteration with thresholding (or nonlinear shrinkage) applied at each iteration step. We prove that this algorithm converges in norm. © 2004 Wiley Periodicals, Inc.
Cited by 4,930 works. doi:10.1002/cpa.20042 Open-access PDF
Authors
References
- Regression Shrinkage and Selection Via the Lasso
- A Wavelet Tour of Signal Processing
- Ideal Spatial Adaptation by Wavelet Shrinkage
- De-noising by soft-thresholding
- Independent Component Analysis
- Atomic Decomposition by Basis Pursuit
- Wavelet Shrinkage: Asymptopia?
- Advances in Electronics and Electron Physics
- Independent Component Analysis
- A Class of Nonharmonic Fourier Series