An alternating direction algorithm for matrix completion with nonnegative factors

Abstract

This paper introduces an algorithm for the nonnegative matrix factorization-and-completion problem, which aims to find nonnegative low-rank matrices X and Y so that the product XY approximates a nonnegative data matrix M whose elements are partially known (to a certain accuracy). This problem aggregates two existing problems: (i) nonnegative matrix factorization where all entries of M are given, and (ii) low-rank matrix completion where nonnegativity is not required. By taking the advantages of both nonnegativity and low-rankness, one can generally obtain superior results than those of just using one of the two properties. We propose to solve the non-convex constrained least-squares problem using an algorithm based on the classical alternating direction augmented Lagrangian method. Preliminary convergence properties of the algorithm and numerical simulation results are presented. Compared to a recent algorithm for nonnegative matrix factorization, the proposed algorithm produces factorizations of similar quality using only about half of the matrix entries. On tasks of recovering incomplete grayscale and hyperspectral images, the proposed algorithm yields overall better qualities than those produced by two recent matrix-completion algorithms that do not exploit nonnegativity.

Keywords

nonnegative matrix factorization / matrix completion / alternating direction method / hyperspectral unmixing

References

[1] Berry M W, Browne M, Langville A N, Pauca V P, Plemmons R J. Algorithms and applications for approximate nonnegative matrix factorization. Comput Statist Data Anal, 2007, 52(1): 155-173

[2] Bertsekas D P, Tsitsiklis J N. Parallel and Distributed Computation: Numerical Methods. Upper Saddle River: Prentice-Hall, Inc, 1989

[3] Biswas P, Lian T C, Wang T C, Ye Y. Semidefinite programming based algorithms for sensor network localization. ACM Trans Sensor Networks, 2006, 2(2): 188-220

[4] Cai J F, Candes E J, Shen Z. A singular value thresholding algorithm for matrix completion export. SIAM J Optim, 2010, 20: 1956-1982

[5] Cand`es E J, Li X, Ma Y, Wright J. Robust principal component analysis? J ACM, 2011, 58(3): 11

[6] Cand`es E J, Recht B. Exact matrix completion via convex optimization. Found Comput Math, 2009, 9(6): 717-772

[7] Cand`es E J, Tao T. The power of convex relaxation: Near-optimal matrix completion. IEEE Trans Inform Theory, 2010, 56(5): 2053-2080

[8] Cichocki A, Morup M, Smaragdis P, Wang W, Zdunek R. Advances in Nonnegative Matrix and Tensor Factorization. Computational Intelligence Neuroscience. New York: Hindawi Publishing Corporation, 2008

[9] Cichocki A, Zdunek R, Phan A H, Amari S. Nonnegative Matrix and Tensor Factorizations—Applications to Exploratory Multiway Data Analysis and Blind Source Separation. Hoboken: John Wiley & Sons, Ltd, 2009

[10] Fazel M. Matrix Rank Minimization with Applications. PhD Thesis, Stanford University. 2002

[11] Gabay D, Mercier B. A dual algorithm for the solution of nonlinear variational problems via finite element approximation. Comput Math Appl, 1976, 2(1): 17-40

[12] Glowinski R, Marrocco A. Sur lapproximation par elements finis dordre un, et la resolution par penalisation-dualite dune classe de problemes de Dirichlet nonlineaires. Rev Francaise dAut Inf Rech Oper, 1975, 41-76

[13] Goldberg D, Nichols D, Oki B M, Terry D. Using collaborative filtering to weave an information tapestry. Commun ACM, 1992, 35(12): 61-70

[14] Goldfarb D, Ma S, Wen Z. Solving low-rank matrix completion problems efficiently. In: Proceedings of 47th Annual Allerton Conference on Communication, Control, and Computing, Monticello, Illinois. 2009

[15] Grippo L, Sciandrone M. On the convergence of the block nonlinear Gauss-Seidel method under convex constraints. Oper Res Lett, 2000, 26(3): 127-136

[16] Hale E T, YinW, Zhang Y. Fixed-point continuation for l1-minimization: methodology and convergence. SIAM J Optim, 2008, 19(3): 1107-1130

[17] Hestenes M R. Multiplier and gradient methods. J Optim Theory Appl, 1969, 4(5): 303-320

[18] Lee D D, Seung H S. Learning the parts of objects by non-negative matrix factorization. Nature, 1999, 401(6755): 788-791

[19] Lee D D, Seung H S. Algorithms for non-negative matrix factorization. Adv Neural Inf Process Syst, 2001, 13: 556-562

[20] Liu Z, Vandenberghe L. Interior-point method for nuclear norm approximation with application to system identification. SIAM J Matrix Anal Appl, 2009, 31(3): 1235-1256

[21] Ma S, Goldfarb D, Chen L. Fixed point and Bregman iterative methods for matrix rank minimization. Math Program, Ser A, 2011, 128(1-2): 321-353

[22] Paatero P. Least squares formulation of robust non-negative factor analysis. Chemometrics Intell Lab Syst, 1997, 37(1): 23-35

[23] Paatero P. The multilinear engine: A table-driven, least squares program for solving multilinear problems, including the n-way parallel factor analysis model. J Comput Graph Statist, 1999, 8(4): 854-888

[24] Paatero P, Tapper U. Positive matrix factorization: A non-negative factor model with optimal utilization of error estimates of data values. Environmetrics, 1994, 5(2): 111-126

[25] Powell M J D. A method for nonlinear constraints in minimization problems. In: Fletcher R, ed. Optimization. New York: Academic Press, 1969, 283-298

[26] Recht B, Fazel M, Parrilo P A. Guaranteed minimum-rank solutions of linear matrix equations via nuclear norm minimization. SIAM Review, 2010, 52(3): 471-501

[27] Rockafellar R T. The multiplier method of Hestenes and Powell applied to convex programming. J Optim Theory Appl, 1973, 12(6): 555-562

[28] Wang Y, Yang J, Yin W, Zhang Y. A new alternating minimization algorithm for total variation image reconstruction. SIAM J Imaging Sci, 2008, 1(3): 248-272

[29] Wen Z, Goldfarb D, Yin W. Alternating direction augmented Lagrangian methods for semidefinite programming. Math ProgramComput, 2010, 2(3-4): 203-230

[30] Wen Z, Yin W, Zhang Y. Solving a low-rank factorization model for matrix completion by a non-linear successive over-relaxation algorithm. Rice Univ CAAM Technical Report TR 10-07, 2010

[31] Yang J, Yuan X. Linearized augmented Lagrangian and alternating direction methods for nuclear norm minimization. Math Comp (to appear)

[32] Yang J, Zhang Y, Yin W. An efficient TVL1 algorithm for deblurring multichannel images corrupted by impulsive noise. SIAM J Sci Comput, 2008, 31: 2842-2865

[33] Yin W, Osher S, Goldfarb D, Darbon J. Bregman iterative algorithms for 1-minimization with applications to compressed sensing. SIAM J Imaging Sci, 2008, 1(1): 143-168

[34] Zhang Y. An alternating direction algorithm for nonnegative matrix factorization. Rice Technical Report TR 10-03, 2010

RIGHTS & PERMISSIONS

2014 Higher Education Press and Springer-Verlag Berlin Heidelberg