A Missing Data Reconstruction Method Using an Accelerated Least-Squares Approximation with Randomized SVD
Abstract
:1. Introduction
- We propose an algorithm that accelerates the missing data reconstruction procedure based on the least-squares method, which can reduce computational time in two aspects. The first one is based on applying the rSVD to obtain an optimal low-dimensional basis used in the least-squares projection. The second aspect arises from applying a greedy point selection procedure that can extract a small subset of important components to reduce the size of the reconstruction problem and, therefore, speed up the computational time.
- We provide numerical experiments that demonstrate the efficiency (in terms of accuracy and speedup) of the proposed method for some standard testing images with different amounts of missing data.
- We perform numerical tests on a sequence of two-dimensional miscible flow images. The proposed method was shown to give 10 to 100 times reduction in reconstruction time with the same order of accuracy. This illustrates the extensibility of the proposed method on other large-scale problems, as well as the applications of video processing.
2. Standard Projection-Based Approach for Data Reconstruction
Algorithm 1 Standard POD-LS approach for approximating missing data |
INPUT: - Complete snapshot set and, dimension - Incomplete data with known entries in and unknown entries in , where . OUTPUT: - Approximation of 1: Create snapshot matrix and let . 2: Construct basis of rank for . 3: Find coefficient vector from using least-squares problem in (1): . 4: Compute the approximation . |
3. Optimal Basis for Data Set
4. Randomized SVD (rSVD)
Algorithm 2 Randomized Singular Value Decomposition (rSVD) |
INPUT: , k where OUTPUT: POD basis: , Singular values: 1: Form 2: Construct random matrix , , where p is a positive integer 3: 4: Set 5: Compute SVD of : 6: Basis: 7: Truncation: , |
5. Accelerated POD-LS Method (Gappy POD)
5.1. Discrete Empirical Interpolation Method (DEIM)
Algorithm 3 Algorithm to create for Interpolation Indices DEIM |
INPUT: linearly independent OUTPUT: and 1: 2: 3: for to m do 4: Solve 5: 6: 7: 8: end for |
5.2. Accelerated POD-LS Method by Using DEIM
Algorithm 4 Accelerated POD-LS method for approximating missing data |
INPUT: - Complete snapshot set and, dimension - Incomplete data with known entries , and unknown entries , OUTPUT: - Approximation: , 1: Create snapshot matrix and let . 2: Compute POD basis of rank for from rSVD in Algorithm 2. 3: Find coefficient vector 3.1 Find indices and from Algorithm 3 by using input from either (i) defined in (8), or (ii) Left singular vectors of using SVD or rSVD 3.2 Solve from (9): . 4: Compute the approximation . |
6. Computational Complexity
7. Numerical Results
7.1. Numerical Test 1
7.2. Numerical Test 2
7.3. Numerical Test 3
8. Conclusions
Author Contributions
Funding
Institutional Review Board Statement
Informed Consent Statement
Data Availability Statement
Acknowledgments
Conflicts of Interest
Abbreviations
POD | Proper Orthogonal Decomposition |
DEIM | Discrete Empirical Interpolation |
LS | Least-squares |
SVD | Singular Value Decomposition |
rSVD | Randomized Singular Value Decomposition |
References
- Zhang, Q.; Yuan, Q.; Zeng, C.; Li, X.; Wei, Y. Missing data reconstruction in remote sensing image with a unified spatial–temporal–spectral deep convolutional neural network. IEEE Trans. Geosci. Remote Sens. 2018, 56, 4274–4288. [Google Scholar] [CrossRef] [Green Version]
- Chai, X.; Gu, H.; Li, F.; Duan, H.; Hu, X.; Lin, K. Deep learning for irregularly and regularly missing data reconstruction. Sci. Rep. 2020, 10, 3302. [Google Scholar]
- Baraldi, P.; Di Maio, F.; Genini, D.; Zio, E. Reconstruction of missing data in multidimensional time series by fuzzy similarity. Appl. Soft Comput. 2015, 26, 1–9. [Google Scholar] [CrossRef]
- Zhou, Y.; Wang, S.; Wu, T.; Feng, L.; Wu, W.; Luo, J.; Zhang, X.; Yan, N. For-backward LSTM-based missing data reconstruction for time-series Landsat images. Gisci. Remote Sens. 2022, 59, 410–430. [Google Scholar] [CrossRef]
- Hafner, J.L.; Deenadhayalan, V.; Rao, K.; Tomlin, J.A. Matrix Methods for Lost Data Reconstruction in Erasure Codes; USENIX Association: Berkeley, CA, USA, 2005; Volume 5, pp. 15–30. [Google Scholar]
- Kreimer, N.; Stanton, A.; Sacchi, M.D. Tensor completion based on nuclear norm minimization for 5D seismic data reconstruction. Geophysics 2013, 78, V273–V284. [Google Scholar] [CrossRef] [Green Version]
- Liu, T.; Wei, H.; Zhang, K. Wind power prediction with missing data using Gaussian process regression and multiple imputation. Appl. Soft Comput. 2018, 71, 905–916. [Google Scholar] [CrossRef]
- Maillo, J.; Ramírez, S.; Triguero, I.; Herrera, F. kNN-IS: An Iterative Spark-based design of the k-Nearest Neighbors classifier for big data. Knowl. Based Syst. 2017, 117, 3–15. [Google Scholar] [CrossRef] [Green Version]
- Liu, Z.-G.; Liu, Y.; Dezert, J.; Pan, Q. Classification of incomplete data based on belief functions and K-nearest neighbors. Knowl.-Based Syst. 2015, 89, 113–125. [Google Scholar] [CrossRef]
- Scholz, M.; Kaplan, F.; Guy, C.L.; Kopka, J.; Selbig, J. Non-linear PCA: A missing data approach. Bioinformatics 2005, 21, 3887–3895. [Google Scholar] [CrossRef] [Green Version]
- Bø, T.H.; Dysvik, B.; Jonassen, I. LSimpute: Accurate estimation of missing values in microarray data with least squares methods. Nucleic Acids Res. 2004, 32, e34. [Google Scholar] [CrossRef] [Green Version]
- Ren, H.; Gao, F.; Jiang, X. Least-squares method for data reconstruction from gradient data in deflectometry. Appl. Opt. 2016, 55, 6052–6059. [Google Scholar] [CrossRef] [Green Version]
- Kaplan, S.T.; Naghizadeh, M.; Sacchi, M.D. Data reconstruction with shot-profile least-squares migration. Geophysics 2010, 75, WB121–WB136. [Google Scholar] [CrossRef]
- Berkooz, G.; Holmes, P.; Lumley, J.L. The proper orthogonal decomposition in the analysis of turbulent flows. Annual Rev. Fluid Mech. 1993, 25, 539–575. [Google Scholar] [CrossRef]
- Lanata, F.; Grosso, A.D. Damage detection and localization for continuous static monitoring of structures using a proper orthogonal decomposition of signals. Smart Mater. Struct. 2006, 15, 1811. [Google Scholar] [CrossRef]
- Schenone, E. Reduced Order Models, Forward and Inverse Problems in Cardiac Electrophysiology. Ph.D. Thesis, Université Pierre et Marie Curie, Paris, France, 2014. [Google Scholar]
- Gurka, R.; Liberzon, A.; Hetsroni, G. POD of vorticity fields: A method for spatial characterization of coherent structures. Int. J. Heat Fluid Flow 2006, 27, 416–423. [Google Scholar] [CrossRef]
- Tsering-Xiao, B.; Xu, Q. Gappy POD-based reconstruction of the temperature field in Tibet. Theor. Appl. Climatol. 2019, 138, 1179–1188. [Google Scholar] [CrossRef]
- Bui-Thanh, T.; Damodaran, M.; Willcox, K. Aerodynamic Data Reconstruction and Inverse Design Using Proper Orthogonal Decomposition. AIAA 2004, 42, 1505–1516. [Google Scholar] [CrossRef] [Green Version]
- Bizon, K.; Continillo, G.; Merola, S.S.; Vaglieco, B.M. Reconstruction of flame kinematics and analysis of cycle variation in a Spark Ignition Engine by means of Proper Orthogonal Decomposition. Comput. Aided Chem. Eng. 2009, 26, 1039–1043. [Google Scholar] [CrossRef]
- Choi, O.; Lee, M.C. Investigation into the combustion instability of synthetic natural gases using high speed flame images and their proper orthogonal decomposition. Int. J. Hydrog. Energy 2016, 41, 20731–20743. [Google Scholar] [CrossRef]
- Bouhoubeiny, E.; Druault, P. Note on the POD-based time interpolation from successive PIV images. Comptes Rendus MéCanique 2009, 337, 776–780. [Google Scholar] [CrossRef] [Green Version]
- Wang, M.; Dutta, D.; Kim, K.; Brigham, J.C. A computationally efficient approach for inverse material characterization combining Gappy {POD} with direct inversion. Comput. Methods Appl. Mech. Eng. 2015, 286, 373–393. [Google Scholar] [CrossRef]
- Lei, J.; Qiu, J.; Liu, S. Dynamic reconstruction algorithm for electrical capacitance tomography based on the proper orthogonal decomposition. Appl. Math. Model. 2015, 39, 6925–6940. [Google Scholar] [CrossRef]
- Sun, X.-H.; Xu, M.-H. Optimal control of water flooding reservoir using proper orthogonal decomposition. J. Comput. Appl. Math. 2017, 320, 120–137. [Google Scholar] [CrossRef]
- Halko, N.; Martinsson, P.; Tropp, J. Finding Structure with Randomness: Probabilistic Algorithms for Constructing Approximate Matrix Decompositions. SIAM Rev. 2011, 53, 217–288. [Google Scholar] [CrossRef]
- Ji, H.; Yu, W.; Li, Y. A Rank Revealing Randomized Singular Value Decomposition (R3SVD) Algorithm for Low-rank Matrix Approximations. arXiv 2016, arXiv:1605.08134. [Google Scholar]
- Zhang, J.; Erway, J.; Hu, X.; Zhang, Q.; Plemmons, R. Randomized SVD Methods in Hyperspectral Imaging. J. Electr. Comput. Eng. 2012, 2012, 2737–2764. [Google Scholar] [CrossRef] [Green Version]
- Ji, H.; Li, Y. GPU Accelerated Randomized Singular Value Decomposition and Its Application in Image Compression. In Proceedings of the MSVESCC, Suffolk, VA, USA, 17 April 2014. [Google Scholar]
- Intawichai, S.; Chaturantabut, S. Missing Image Data Reconstruction Based on Least-Squares Approach with Randomized SVD. In Proceedings of the International Conference on Intelligent Computing & Optimization, Koh Samui, Thailand, 17–18 December 2020; Springer: Berlin/Heidelberg, Germany, 2020; pp. 1059–1070. [Google Scholar]
- Yu, W.; Gu, Y.; Li, Y. Efficient randomized algorithms for the fixed-precision low-rank matrix approximation. Siam J. Matrix Anal. Appl. 2018, 39, 1339–1359. [Google Scholar] [CrossRef]
- Voronin, S.; Martinsson, P.G. RSVDPACK: An implementation of randomized algorithms for computing the singular value, interpolative, and CUR decompositions of matrices on multi-core and GPU architectures. arXiv 2015, arXiv:1502.05366. [Google Scholar]
- Musco, C.; Musco, C. Randomized Block Krylov Methods for Stronger and Faster Approximate Singular Value Decomposition. In Proceedings of the 28th International Conference on Neural Information Processing Systems, Montreal, QC, Canada, 7–12 December 2015; MIT Press: Cambridge, MA, USA, 2015; Volume 1, pp. 1396–1404. [Google Scholar]
- Li, H.; Linderman, G.C.; Szlam, A.; Stanton, K.P.; Kluger, Y.; Tygert, M. Algorithm 971: An implementation of a randomized algorithm for principal component analysis. ACM Trans. Math. Softw. 2017, 43, 1–14. [Google Scholar] [CrossRef]
- Chaturantabut, S.; Sorensen, D.C. Nonlinear Model Reduction via Discrete Empirical Interpolation. SIAM J. Sci. Comput. 2010, 32, 2737–2764. [Google Scholar] [CrossRef]
- Wirtz, D.; Sorensen, D.C.; Haasdonk, B. A posteriori error estimation for DEIM reduced nonlinear dynamical systems. Siam J. Sci. Comput. 2014, 36, A311–A338. [Google Scholar] [CrossRef] [Green Version]
- Chaturantabut, S. Stabilized model reduction for nonlinear dynamical systems through a contractivity-preserving framework. Int. J. Appl. Math. Comput. Sci. 2020, 30, 615–628. [Google Scholar]
- Ştefănescu, R.; Navon, I.M. POD/DEIM nonlinear model order reduction of an ADI implicit shallow water equations model. J. Comput. Phys. 2013, 237, 95–114. [Google Scholar] [CrossRef] [Green Version]
- Stefanescu, R.; Sandu, A.; Navon, I.M. POD/DEIM Strategies for reduced data assimilation systems. J. Comput. Phys. 2014, 295, 569–595. [Google Scholar]
- Feng, Z.; Soulaimani, A. Reduced Order Modelling Based on POD Method for 3D Nonlinear Aeroelasticity. In Proceedings of the 18th IASTED International Conference on Modelling and Simulation, Montreal, QC, Canada, 30 May–1 June 2007; ACTA Press: Anaheim, CA, USA, 2007; pp. 489–494. [Google Scholar]
- Ploymaklam, N.; Chaturantabut, S. Reduced-Order Modeling of a Local Discontinuous Galerkin Method for Burgers-Poisson Equations. Thai J. Math. 2020, 18, 2053–2069. [Google Scholar]
- Xiao, D.; Fang, F.; Buchan, A.G.; Pain, C.C.; Navon, I.M.; Du, J.; Hu, G. Non-linear model reduction for the Navier–Stokes equations using residual DEIM method. J. Comput. Phys. 2014, 263, 1–18. [Google Scholar] [CrossRef] [Green Version]
- Chaturantabut, S. Accelerated POD least-squares approach for missing data reconstruction. In Proceedings of the 17th International Conference on Mathematical Methods in Science and Engineering, Beijing, China, 21–22 May 2017; pp. 563–575. [Google Scholar]
- Eckart, C.; Young, G. The approximation of one matrix by another of lower rank. Psychometrika 1936, 1, 211–218. [Google Scholar] [CrossRef]
- Volkwein, S. Proper Orthogonal Decomposition: Theory and Reduced-Order Modelling; Lecture Notes; University of Konstanz: Konstanz, Germany, 2013. [Google Scholar]
- Gu, M.; Eisenstat, S.C. Efficient Algorithms for Computing a Strong Rank-Revealing QR Factorization. Siam J. Sci. Comput. 1996, 17, 848–869. [Google Scholar] [CrossRef] [Green Version]
- Lawson, C.L.; Hanson, R.J. Solving Least Squares Problems; SIAM: Philadelphia, PA, USA, 1995. [Google Scholar]
- Golub, G. Numerical methods for solving linear least squares problems. Numer. Math. 1965, 7, 206–216. [Google Scholar] [CrossRef]
- Lee, D.Q. Numerically Efficient Methods for Solving Least Squares Problems; Pennsylvania State University: State College, PA, USA, 2012. [Google Scholar]
- Gu, Z.; Lin, W.; Lee, B.S.; Lau, C.T.; Paul, M. Two dimensional singular value decomposition (2D-SVD) based video coding. In Proceedings of the 2010 IEEE International Conference on Image Processing, Haikou, China, 11–12 November 2010; pp. 181–184. [Google Scholar]
- Gutiérrez, P.A.; Hervás, C.; Carbonero, M.; Fernández, J.C. Combined projection and kernel basis functions for classification in evolutionary neural networks. Neurocomputing 2009, 72, 2731–2742. [Google Scholar] [CrossRef]
- Diener, J.; Rodriguez, M.; Baboud, L.; Reveret, L. Wind projection basis for real-time animation of trees. In Computer Graphics Forum; Wiley Online Library: Hoboken, NJ, USA, 2009; Volume 28, pp. 533–540. [Google Scholar]
- Liu, X.; Liu, G.; Tai, K.; Lam, K. Radial point interpolation collocation method (RPICM) for the solution of nonlinear Poisson problems. Comput. Mech. 2005, 36, 298–306. [Google Scholar] [CrossRef]
- Weickert, J.; Welk, M. Tensor field interpolation with PDEs. In Visualization and Processing of Rensor Fields; Springer: Berlin/Heidelberg, Germany, 2006; pp. 315–325. [Google Scholar]
- Nguyen, N.C.; Patera, A.T.; Peraire, J. A ‘best points’ interpolation method for efficient approximation of parametrized functions. Int. J. Numer. Methods Eng. 2008, 73, 521–543. [Google Scholar] [CrossRef]
- Groza, G.; Pop, N. Approximate solution of multipoint boundary value problems for linear differential equations by polynomial functions. J. Differ. Equ. Appl. 2008, 14, 1289–1309. [Google Scholar] [CrossRef]
Method | dim POD | dim DEIM | Relative Error | CPU Time (Scaled) |
---|---|---|---|---|
Standard POD-LS | k = 10 | - | 1 | |
Accelerated POD-LS | k = 10 | m = 10 | ||
Method | dim POD | dim DEIM | Relative Error | CPU Time (Scaled) |
---|---|---|---|---|
Standard POD-LS | k = 50 | - | 1 | |
Accelerated POD-LS | ||||
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |
© 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Share and Cite
Intawichai, S.; Chaturantabut, S. A Missing Data Reconstruction Method Using an Accelerated Least-Squares Approximation with Randomized SVD. Algorithms 2022, 15, 190. https://doi.org/10.3390/a15060190
Intawichai S, Chaturantabut S. A Missing Data Reconstruction Method Using an Accelerated Least-Squares Approximation with Randomized SVD. Algorithms. 2022; 15(6):190. https://doi.org/10.3390/a15060190
Chicago/Turabian StyleIntawichai, Siriwan, and Saifon Chaturantabut. 2022. "A Missing Data Reconstruction Method Using an Accelerated Least-Squares Approximation with Randomized SVD" Algorithms 15, no. 6: 190. https://doi.org/10.3390/a15060190
APA StyleIntawichai, S., & Chaturantabut, S. (2022). A Missing Data Reconstruction Method Using an Accelerated Least-Squares Approximation with Randomized SVD. Algorithms, 15(6), 190. https://doi.org/10.3390/a15060190