Abstract
Sparse representation (SR) is a fundamental component of linear representation techniques and plays a crucial role in signal processing, machine learning, and computer vision. Most parallel methods for solving sparse representations rely on the alternating direction method of multipliers (ADMM). However, the classical 2-block ADMM or N-block ADMM often suffer from three problems: (1) solving the subproblem requires solving a linear system, (2) unsuitable sparse data structure for parallelization, and (3) unsatisfactory parallel efficiency and scalability performance. In this paper, we propose a parallel ADMM-based algorithm called block splitting proximal ADMM (BSPADMM). First, BSPADMM organizes the sparse signals in the compressed sparse columns (CSC) format, and each processor deals with them independently. Second, BSPADMM designs the proximal term that avoids solving a linear system of the subproblem during iterations. Its advantage is that the BSPADMM computes the subproblem by using sparse matrix–vector multiplication, without communication between processors. Third, each processor updates the size asynchronously, which eliminates the synchronization effort of adjusting the step size between processes. Thus, the communication overhead can be naturally reduced. Our experimental results on three datasets of varying scales show that BSPADMM outperforms state-of-the-art ADMM techniques, including the adaptive relaxed ADMM (ARADMM) and N-block ADMM, in terms of computing time and parallel efficiency. BSPADMM runs 1.64 times faster than the N-block ADMM, and the ratio grows to 8.27 times as the dataset size doubles. More importantly, the parallel efficiency of BSPADMM remains above \(70\%\) as the number of processors grows to 10,000, demonstrating strong scalability.
Similar content being viewed by others
References
Noise modeling and representation based classification methods for face recognition. Neurocomputing. 148 , 420–429 (2015)
Edoardo Amaldi, A., Viggo Kann, B.: On the approximability of minimizing nonzero variables or unsatisfied relations in linear systems. Theoretical Computer Science. 209(1–2), 237–260 (1998)
Bao, C., Ji, H., Quan, Y., Shen, Z.: L0 Norm based dictionary learning by proximal methods with global convergence. In 2014 IEEE Conference on Computer Vision and Pattern Recognition. pp. 3858–3865 (2014)
Bao, C., Wu, Y., Ling, H., Ji, H.: Real time robust L1 tracker using accelerated proximal gradient approach. In 2012 IEEE Conference on Computer Vision and Pattern Recognition. pp. 1830–1837 (2012)
Boyd, S., Parikh, N., Chu, E., Peleato, B., Eckstein, J.: Distributed optimization and statistical learning via the alternating direction method of multipliers. Foundations and trends in machine learning. 3, (1): 1–122 (2011)
Cai, Xingju, Han, Deren, Yuan, Xiaoming: On the convergence of the direct extension of ADMM for three-block separable convex minimization models with one strongly convex function. Comput Optim Appl. 66(1), 39–73 (2017)
NVIDIA Corporation. NVIDAM CuSolver. NVIDIA. https://developer.nvidia.com/cusolver (2015)
Dai, Y.-H., Hager, W.W., Schittkowski, K., Zhang, H.: The cyclic Barzilai-C̈Borwein method for unconstrained optimization. IMA J. Numer. Anal. 26(3), 604–627 (2006)
Davis, Damek: A three-operator splitting scheme and its optimization applications. Set-Valued Var Anal 25(2017), 829–858 (2017)
Deng, X., Liu, F., Huang, F.: Linear convergence rate of splitting algorithms for multi-block constrained convex minimizations. IEEE Access. 8, 120694–120700 (2020)
For most large underdetermined systems of linear equations the minimal \(l_{1}\)-norm solution is also the sparsest solution. Commun Pure Appl Math. 59(6), 797–829 (2006)
Donoho, D.L., Tsaig, Y.: Fast solution of \(\ell _{1}\) -norm minimization problems when the solution may be sparse. IEEE Trans Inf Theory. 54(11), 4789–4812 (2008)
Elgabli, A., Elghariani, A., Aggarwal, V., Bennis, M., Bell, M.: A proximal Jacobian ADMM approach for fast massive MIMO signal detection in low-latency communications. IEEE. pp. 1–6 (2019)
Field, J.D.: Relations between the statistics of natural images and the response properties of cortical cells. J Opt Soc Am A-Opt Image Sci Vis. 4(12), 2379–2394 (1987)
Gropp, W., Smith, B., McInnes, L.C.: PETSC 2.0. Portable Extensible Toolkit for Scientific Computation. United States (1995)
Han, D., Yuan, X.: A note on the alternating direction method of multipliers. J Optim Theory Appl. 155(1), 227–238 (2012)
He, B., Yuan, X..: On non-ergodic convergence rate of Douglas—Rachford alternating direction method of multipliers. Numer. Math. 130, (3): 567C̈577 (2015)
Hong, M., Luo, Z-Q.: On the linear convergence of the alternating direction method of multipliers. Math. Program. 162, 1C̈2 (2017), 165C̈199 (2017)
Hu, W., Qin, X., Jiang, Q., Chen, J., An, H., Jia, W., Yang, C., Wang, L., Yang, C., Lin, L.: High performance computing of DGDFT for tens of thousands of atoms using millions of cores on Sunway TaihuLight. Sci Bull. 66(2), 111–119 (2021)
Huang, K., Aviyente, S.: Sparse Representation for signal classification. In Proceedings of the 19th International Conference on Neural Information Processing Systems (Canada) (NIPS’06). MIT Press, Cambridge. pp. 609C̈616 (2006)
Hubel, D.H., Wiesel, T.N.: Receptive fields of single neurones in the cat’s striate cortex. J Physiol. 148(3), 574–591 (1959)
Jenatton, R., Mairal, J., Obozinski, G., Bach, F..: Proximal methods for sparse hierarchical dictionary learning. Proceedings of the International Conference on Machine Learning (ICML). pp. 487–494 (2010)
Jin, J.W., Wen, S.: An algorithm twisted from generalized ADMM for multi-block separable convex minimization models. J. Comput. Appl. Math. 309(2017), 342–358 (2017)
Krause, A., Cevher, V..: Submodular dictionary selection for sparse representation. In Proceedings of the 27th International Conference on International Conference on Machine Learning (Haifa, Israel) (ICML’10). Omnipress, Madison. pp. 567C̈574. 9781605589077 (2010)
LAPACK.: LAPACK linear system solver. LAPACK. https://netlib.org/lapack/ (2000)
Lazebnik, S., Schmid, C., Ponce, J.: A sparse texture representation using affine-invariant regions. In 2003 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2003. Proceedings. 2, 109–123 (2003)
Li, M., Sun, D., Toh, K.-C.: A Convergent 3-block semi-proximal ADMM for convex minimization problems with one strongly convex block. Asia-Pacific J Operat Res. 32, 1550024 (2015)
Lin, T., Ma, S., Zhang, S.: On the global linear convergence of the ADMM with multiblock variables. SIAM J Optim. 25(3), 1478–1497 (2015)
Lin, T., Ma, S., Zhang, S.: Iteration complexity analysis of multi-block ADMM for a family of convex minimization without strong convexity. J Sci Comput. 69(1), 52–81 (2016)
Lin, T., Ma, S., Zhang, S.: Iteration complexity analysis of multi-block ADMM for a family of convex minimization without strong convexity. J. Sci. Comput. 69(1): 52C̈81 (2016b)
Liu, H., Song, B., Qin, H., Qiu, Z.: An adaptive-ADMM algorithm with support and signal value detection for compressed sensing. IEEE Signal Process Lett. 20(4), 315–318 (2013). https://doi.org/10.1109/LSP.2013.2245893
Nesterov, Y.: Excessive gap technique in nonsmooth convex minimization. SIAM J. Optim. 16(1): 235C̈249. 1052-6234 (2005)
Patel, VM., Chellappa, R.: Sparse representations, compressive sensing and dictionaries for pattern recognition. In The First Asian Conference on Pattern Recognition. pp. 325–329 (2011)
Peng, G.-J.: Adaptive ADMM for dictionary learning in convolutional sparse representation. IEEE Trans Image Process. 28(7), 3408–3422 (2019). https://doi.org/10.1109/TIP.2019.2896541
Plumbley, MD.: Recovery of sparse representations by polytope faces Pursuit. In Proceedings of the 6th International Conference on Independent Component Analysis and Blind Signal Separation (Charleston, SC) (ICA’06). Springer-Verlag, Berlin, Heidelberg. pp. 206C̈213. 3540326308 (2006)
Shen, Y., Zuo, Y., Yu, A.: A partially proximal S-ADMM for separable convex optimization with linear constraints. Appl. Numer. Math. 160(2021), 65–83 (2021)
Shurong, Z.: Selection of components and degrees of smoothing via lasso in high dimensional nonparametric additive models. Comput Stat Data Anal 53(1), 164–175 (2008)
Sun, H, Wang, J, Deng, T: On the global and linear convergence of direct extension of ADMM for 3-block separable convex minimization models. J Inequal Appli. pp. 1–14 (2016)
Sun, R., Luo, Z.-Q., Ye, Y.: On the efficiency of random permutation for ADMM and coordinate descent. Math Operat Res. 45, 1–14 (2020)
Tibshirani, R.: Regression shrinkage and selection via the lasso: a retrospective. J R Stat Soc: Series B (Stat Methodol). 73(3), 267–288 (2011)
Tropp, J.A.: Greed is good: algorithmic results for sparse approximation. IEEE Trans Inf Theory 50(10), 2231–2242 (2004)
Wang, J., Yang, J., Yu, K., Lv, F., Huang, T., Gong, Y.: Locality-constrained linear coding for image classification. In 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition. pp. 3360–3367. (2010) https://doi.org/10.1109/CVPR.2010.5540018
Wright, SJ., Nowak, RD., Figueiredo, MAT.: Sparse reconstruction by separable approximation. Trans. Sig. Proc. 57(7): 2479C̈2493 (2009)
Xu, Z., Figueiredo, MT., Yuan, X., Studer, C., Goldstein, T.: Adaptive relaxed ADMM: convergence theory and practical implementation. In 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). pp. 7234–7243 (2017)
Yang, A.Y., Ma, Y., Wright, J., Ganesh, A., Sastry, S.: Robust face recognition via sparse representation. IEEE Trans Pattern Anal Mach Intell. 31, 210–227 (2009)
Yang, J., Wright, J., Huang, T., Ma, Y.: Image super-resolution as sparse representation of raw image patches. In 2008 IEEE Conference on Computer Vision and Pattern Recognition. pp. 1–8 (2008)
Zhang, Z., Xu, Y., Yang, J., Li, X., Zhang, D.: A survey of sparse representation: algorithms and applications. IEEE Access. 3, 490–530 (2015). https://doi.org/10.1109/ACCESS.2015.2430359
Acknowledgements
This material is based upon work supported by the National Science Foundationhttp://dx.doi.org/10.13039/100000001 under Grant No.nnnnnnn and Grant No.mmmmmmm. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author and do not necessarily reflect the views of the National Science Foundation.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors declared that they have no conflicts of interest to this work. We declare that we do not have any commercial or associative interest that represents a conflict of interest in connection with the work submitted.
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
Chen, Y., Pan, J., Han, Z. et al. BSPADMM: block splitting proximal ADMM for sparse representation with strong scalability. CCF Trans. HPC 6, 3–16 (2024). https://doi.org/10.1007/s42514-023-00164-w
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s42514-023-00164-w