Abstract
We propose a new algorithmic framework for constrained compressed sensing models that admit nonconvex sparsity-inducing regularizers including the log-penalty function as objectives, and nonconvex loss functions such as the Cauchy loss function and the Tukey biweight loss function in the constraint. Our framework employs iteratively reweighted \(\ell _1\) and \(\ell _2\) schemes to construct subproblems that can be efficiently solved by well-developed solvers for basis pursuit denoising such as SPGL1 by van den Berg and Friedlander (SIAM J Sci Comput 31:890-912, 2008). We propose a new termination criterion for the subproblem solvers that allows them to return an infeasible solution, with a suitably constructed feasible point satisfying a descent condition. The feasible point construction step is the key for establishing the well-definedness of our proposed algorithm, and we also prove that any accumulation point of this sequence of feasible points is a stationary point of the constrained compressed sensing model, under suitable assumptions. Finally, we compare numerically our algorithm (with subproblems solved by SPGL1 or the alternating direction method of multipliers) against the SCP\(_\textrm{ls}\) in Yu et al. (SIAM J Optim 31: 2024-2054, 2021) on solving constrained compressed sensing models with the log-penalty function as the objective and the Cauchy loss function in the constraint, for badly scaled measurement matrices. Our computational results show that our approaches return solutions with better recovery errors, and are always faster.
Similar content being viewed by others
Data availibility
The codes for generating the random data and implementing the algorithms in the numerical section are available at https://www.polyu.edu.hk/ama/profile/pong/, the webpage of the corresponding author.
Notes
As we shall see later, their continuities are important in establishing convergence of our proposed algorithm; see the proof of Theorem 3.1(v) below.
Note that \({{\overline{\phi }}} > 0\) because \(\lim _{t\downarrow 0}\phi '(t) > 0\) and \(\phi (0)= 0\) in view of Assumption 1.1(ii).
One possible choice is to set \(\rho ={\bar{L}}\beta \), where \({\bar{L}}:=\max \limits _{i}\{\phi '_+((a^T_ix^k - b_i)^2)\}\lambda _{\max }(A^TA)\). With this choice of \({{\bar{L}}}\), one can see from the definition of \({\bar{A}}\) in (4.1) that \({\bar{L}}\ge {\lambda _{\max }({\bar{A}}^T{\bar{A}})}\), and thus \(\rho I-\beta {\bar{A}}^T{\bar{A}}=\beta ({\bar{L}}I-{\bar{A}}^T{\bar{A}})\succeq {0}\).
This means that while the \({{\tilde{\varphi }}}_l\) and \({{\tilde{\varphi }}}_l'\) will be generated as in the defaulting setting in SPGL1, we may need to run the SPG method to approximately solve (4.10) for more iterations to obtain a candidate that verifies our inexact criteria (3.6), (3.7) and (3.8). For the validity of our subsequent arguments, we have to emphasize that this more “accurate" solution will not be used to construct \({{\tilde{\varphi }}}_l\) and \({{\tilde{\varphi }}}_l'\).
The nonemptiness follows in particular from \(\Vert {{\bar{b}}}\Vert > 0\); see the discussion following (4.8).
In our numerical experiment, \(L=\lambda _{\max }(AA^{T})\) is computed using the MATLAB commands: if m > 2000 opts.issym = 1; L = eigs(A*A’,1,’LM’,opts); else L = norm(A*A’); end
It is indeed not known whether \(\lim _{k\rightarrow \infty }\Vert x^{k+1}-x^k\Vert =0\) for the sequence \(\{x^k\}\) generated by these algorithms. We use this criterion as a heuristic and it appears to work well.
The codes were downloaded from https://github.com/mpf/spgl1.
Nevertheless, there is no guarantee that the \(\{x^k\}\) thus generated will cluster at a stationary point of (5.1). We include this version in our experiment as a demonstration of how our framework can be used when only a black-box subproblem solver is available.
For fair comparison, we report the total number of inner iterations for \(\textbf{IR}^{\ell _1}_{\ell _2}\) \(_\text {ADMM}\) and \(v\textbf{IR}^{\ell _1}_{\ell _2}\) \(_\text {SPGL1}\), i.e., the total number of iterations used by the subproblem solvers to solve the subproblems.
References
Attouch, H., Bolte, J.: On the convergence of the proximal algorithm for nonsmooth functions involving analytic features. Math. Program. 116, 5–16 (2009)
Attouch, H., Bolte, J., Redont, P., Soubeyran, A.: Proximal alternating minimization and projection methods for nonconvex problems: an approach based on the Kurdyka-Łojasiewicz inequality. Math. Oper. Res. 35, 438–457 (2010)
Attouch, H., Bolte, J., Svaiter, B.F.: Convergence of descent methods for semi-algebraic and tame problems: proximal algorithms, forward-backward splitting, and regularized Gauss-Seidel methods. Math. Program. 137, 91–129 (2013)
Aravkin, A.Y., Burke, J.V., Drusvyatskiy, D., Friedlander, M.P., Roy, S.: Level-set methods for convex optimization. Math. Program. 174, 359–390 (2019)
Barron, J. T.: A general and adaptive robust loss function. In: IEEE/CVF Conf. Comput. Vis. Pattern Recognit. pp. 4331–4339 (2019)
Beck, A.: First-order Methods in Optimization. SIAM (2017)
Beck, A., Teboulle, M.: A fast iterative shrinkage-thresholding algorithm for linear inverse problems. SIAM J. Imaging Sci. 2, 183–202 (2009)
Becker, S.R., Candès, E.J., Grant, M.C.: Templates for convex cone problems with applications to sparse signal recovery. Math. Program. Comput. 3, 165–218 (2011)
Birgin, E.G., Martínez, J.M., Raydan, M.: Nonmonotone spectral projected gradient methods on convex sets. SIAM J. Optim. 10, 1196–1211 (2000)
Bolte, J., Pauwels, E.: Majorization-minimization procedures and convergence of SQP methods for semi-algebraic and tame programs. Math. Oper. Res. 41, 442–465 (2016)
Boyd, S., Parikh, N., Chu, E., Peleato, B., Eckstein, J.: Distributed optimization and statistical learning via the alternating direction method of multipliers. Found. Trends Mach. Learn. 3, 1–122 (2010)
Candès, E.J., Tao, T.: Decoding by linear programming. IEEE Trans. Inf. Theory 51, 4203–4215 (2005)
Candès, E.J., Wakin, M.B., Boyd, S.P.: Enhancing sparsity by reweighted \(\ell _1\) minimization. J. Fourier Anal. Appl. 14, 877–905 (2008)
Carrillo, R.E., Barner, K.E., Aysal, T.C.: Robust sampling and reconstruction methods for sparse signals in the presence of impulsive noise. IEEE J. Sel. Top. Signal Process. 4, 392–408 (2010)
Carrillo, R.E., Ramirez, A.B., Arce, G.R., Barner, K.E., Sadler, B.M.: Robust compressive sensing of sparse signals: a review. EURASIP J. Adv. Signal Process. 108, 1–17 (2016)
Charbonnier, P., Blanc-Féraud, L., Aubert, G., Barlaud, M.: Deterministic edge-preserving regularization in computed imaging. IEEE Trans. Image Process. 6, 298–311 (1997)
Chartrand, R.: Exact reconstruction of sparse signals via nonconvex minimization. IEEE Signal Process. Lett. 14, 707–710 (2007)
Chartrand, R., Yin, W.: Iteratively reweighted algorithms for compressive sensing. In: IEEE Int. Conf. Acoust. Speech Signal Process. pp. 3869–3872 (2008)
Combettes, P.L., Wajs, V.R.: Signal recovery by proximal forward-backward splitting. Multiscale Model. Simul. 4, 1168–1200 (2005)
Dennis, J.E., Jr., Welsch, R.E.: Techniques for nonlinear least squares and robust regression. Commun. Stat. Simul. Comput. 7, 345–359 (1978)
Fazel, M., Pong, T.K., Sun, D., Tseng, P.: Hankel matrix rank minimization with applications to system identification and realization. SIAM J. Matrix Anal. Appl. 34, 946–977 (2013)
Foucart, S., Rauhut, H.: A Mathematical Introduction to Compressive Sensing. Springer, New York (2013)
Geman, S., McClure, D. E.: Bayesian image analysis: An application to single photon emission tomography. Proc. Stat. Comput. Sect. Amer. Statist. Assoc. pp. 12–18 (1985)
Hiriart-Urruty, J.-B., Lemaréchal, C.: Fundamentals of Convex Analysis. Springer, New York (2001)
Huber, P.J.: Robust estimation of a location parameter. Ann. Math. Statist. 35, 73–101 (1964)
Kassam, S.A., Poor, H.V.: Robust techniques for signal processing: a survey. Proc. IEEE 73, 433–481 (1985)
Lange, K.: MM Optimization Algorithms. SIAM (2016)
Le Thi, H.A., Tao, P.D.: DC programming and DCA: thirty years of developments. Math. Program. 169, 5–68 (2018)
Lu, Z.: Iterative reweighted minimization methods for \(l_p\) regularized unconstrained nonlinear programming. Math. Program. 147, 277–307 (2014)
Lu, Z., Zhang, Y.: An augmented Lagrangian approach for sparse principal component analysis. Math. Program. 135, 149–193 (2012)
Mosteller, F., Tukey, J.W.: Data Analysis and Regression: A Second Course in Statistics. Addison-Wesley, Sydney (1977)
Nesterov, Y.: A method for solving a convex programming problem with convergence rate \(O(1/k^2)\). Soviet Math. Dokl. 27, 372–376 (1983)
Nesterov, Y.: Introductory Lectures on Convex Optimization: A Basic Course. Kluwer Academic Publishers, Boston (2004)
Nesterov, Y.: Smooth minimization of non-smooth functions. Math. Program. 103, 127–152 (2005)
Nikolova, M., Ng, M.K., Zhang, S., Ching, W.-K.: Efficient reconstruction of piecewise constant images using nonsmooth nonconvex minimization. SIAM J. Imaging Sci. 1, 2–25 (2008)
Riani, M., Cerioli, A., Atkinson, A.C., Perrotta, D.: Monitoring robust regression. Electron. J. Statist. 8, 646–677 (2014)
Rockafellar, R.T.: Convex Analysis. Princeton University Press, Princeton, NJ (1970)
Rockafellar, R.T., Wets, R.J.-B.: Variational Analysis. Springer, Berlin (1998)
Tseng, P., Yun, S.: A coordinate gradient descent method for nonsmooth separable minimization. Math. Program. 117, 387–423 (2009)
van den Berg, E., Friedlander, M.P.: Probing the Pareto frontier for basis pursuit solutions. SIAM J. Sci. Comput. 31, 890–912 (2008)
Villa, S., Salzo, S., Baldassarre, L., Verri, A.: Accelerated and inexact forward-backward algorithms. SIAM J. Optim. 23, 1607–1633 (2013)
Wang, H., Zhang, F., Shi, Y., Hu, Y.: Nonconvex and nonsmooth sparse optimization via adaptively iterative reweighted methods. J. Glob. Optim. 81, 717–748 (2021)
Yang, J., Zhang, Y.: Alternating direction algorithms for \(\ell _1\)-problems in compressive sensing. SIAM J. Sci. Comput. 33, 250–278 (2011)
Yang, L., Toh, K.-C.: Bregman proximal point algorithm revisited: A new inexact version and its inertial variant. SIAM J. Optim. 32, 1523–1554 (2022)
Yang, L., Toh, K.-C.: An inexact Bregman proximal gradient method and its inertial variant. Preprint (2021). Available at arXiv:https://arxiv.org/abs/2109.05690
Yang, X., Wang, J., Wang, H.: Towards an efficient approach for the nonconvex \(\ell _p\) ball projection: algorithm and analysis. J. Mach. Learn. Res. 23, 1–31 (2022)
Yu, P., Pong, T.K.: Iteratively reweighted \(\ell _1\) algorithms with extrapolation. Comput. Optim. Appl. 73, 353–386 (2019)
Yu, P., Pong, T.K., Lu, Z.: Convergence rate analysis of a sequential convex programming method with line search for a class of constrained difference-of-convex optimization problems. SIAM J. Optim. 31, 2024–2054 (2021)
Zoubir, A.M., Koivunen, V., Chakhchoukh, Y., Muma, M.: Robust estimation in signal processing: a tutorial-style treatment of fundamental concepts. IEEE Signal Process. Mag. 29, 61–80 (2012)
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflicts of interest
The corresponding author is an editorial board member of this journal.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Ting Kei Pong was supported in part by Hong Kong Research Grants Council PolyU153000/20p.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Sun, S., Pong, T.K. Doubly iteratively reweighted algorithm for constrained compressed sensing models. Comput Optim Appl 85, 583–619 (2023). https://doi.org/10.1007/s10589-023-00468-1
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10589-023-00468-1