A Primal Dual Active Set Algorithm With Continuation for Compressed Sensing

被引:38
作者
Fan, Qibin [1 ]
Jiao, Yuling [1 ]
Lu, Xiliang [1 ]
机构
[1] Wuhan Univ, Sch Math & Stat, Wuhan 430072, Peoples R China
基金
美国国家科学基金会;
关键词
Compressive sensing; l(1) regularization; primal dual active set method; continuation; modified discrepancy principle; Bayesian information criterion; CONSTRAINED MINIMIZATION; THRESHOLDING ALGORITHM; SIGNAL RECOVERY; NEWTON METHODS; SELECTION; RECONSTRUCTION; SHRINKAGE; CONVERGENCE; SEMISMOOTH; SPARSITY;
D O I
10.1109/TSP.2014.2362880
中图分类号
TM [电工技术]; TN [电子技术、通信技术];
学科分类号
0808 ; 0809 ;
摘要
The success of compressed sensing relies essentially on the ability to efficiently find an approximately sparse solution to an under-determined linear system. In this paper, we developed an efficient algorithm for the sparsity promoting l(1)-regularized least squares problem by coupling the primal dual active set strategy with a continuation technique (on the regularization parameter). In the active set strategy, we first determine the active set from primal and dual variables, and then update the primal and dual variables by solving a low-dimensional least square problem on the active set, which makes the algorithm very efficient. The continuation technique globalizes the convergence of the algorithm, with provable global convergence under restricted isometry property (RIP). Further, we adopt two alternative methods, i.e., a modified discrepancy principle and a Bayesian information criterion, to choose the regularization parameter automatically. Numerical experiments indicate that our algorithm is very competitive with state-of-the-art algorithms in terms of accuracy and efficiency, without a priori information about the regularization parameter.
引用
收藏
页码:6276 / 6285
页数:10
相关论文
共 51 条
  • [11] Iteratively Reweighted Least Squares Minimization for Sparse Recovery
    Daubechies, Ingrid
    Devore, Ronald
    Fornasier, Massimo
    Guentuerk, C. Sinan
    [J]. COMMUNICATIONS ON PURE AND APPLIED MATHEMATICS, 2010, 63 (01) : 1 - 38
  • [12] Analysis of Orthogonal Matching Pursuit Using the Restricted Isometry Property
    Davenport, Mark A.
    Wakin, Michael B.
    [J]. IEEE TRANSACTIONS ON INFORMATION THEORY, 2010, 56 (09) : 4395 - 4401
  • [13] Donoho D. L., 2006, FAST SOLUTION L1 NOR
  • [14] Compressed sensing
    Donoho, DL
    [J]. IEEE TRANSACTIONS ON INFORMATION THEORY, 2006, 52 (04) : 1289 - 1306
  • [15] Adapting to unknown smoothness via wavelet shrinkage
    Donoho, DL
    Johnstone, IM
    [J]. JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION, 1995, 90 (432) : 1200 - 1224
  • [16] Least angle regression - Rejoinder
    Efron, B
    Hastie, T
    Johnstone, I
    Tibshirani, R
    [J]. ANNALS OF STATISTICS, 2004, 32 (02) : 494 - 499
  • [17] Engl H. W., 1996, REGULARIZATION INVER
  • [18] Gradient Projection for Sparse Reconstruction: Application to Compressed Sensing and Other Inverse Problems
    Figueiredo, Mario A. T.
    Nowak, Robert D.
    Wright, Stephen J.
    [J]. IEEE JOURNAL OF SELECTED TOPICS IN SIGNAL PROCESSING, 2007, 1 (04) : 586 - 597
  • [19] Golub G.H., 1996, Matrix computations, V3
  • [20] Necessary and Sufficient Conditions for Linear Convergence of l1-Regularization
    Grasmair, Markus
    Haltmeier, Markus
    Scherzer, Otmar
    [J]. COMMUNICATIONS ON PURE AND APPLIED MATHEMATICS, 2011, 64 (02) : 161 - 182