Nonnegative tensor factorizations using an alternating direction method

Abstract

The nonnegative tensor (matrix) factorization finds more and more applications in various disciplines including machine learning, data mining, and blind source separation, etc. In computation, the optimization problem involved is solved by alternatively minimizing one factor while the others are fixed. To solve the subproblem efficiently, we first exploit a variable regularization term which makes the subproblem far from ill-condition. Second, an augmented Lagrangian alternating direction method is employed to solve this convex and well-conditioned regularized subproblem, and two accelerating skills are also implemented. Some preliminary numerical experiments are performed to show the improvements of the new method.

Keywords

Nonnegative matrix factorization / nonnegative tensor factorization / nonnegative least squares / alternating direction method

References

[1] Acar E., Dunlavy D. M., Kolda T. G.. A scalable optimization approach for fitting canonical tensor decompositions. J Chemometrics, 2011, 25(2): 67-86

[2] Bader B. W., Kolda T. G.. Algorithm 862: MATLAB tensor classes for fast algorithm prototyping. ACM Trans Math Software, 2006, 32(4): 635-653

[3] Bader B. W., Kolda T. G.. Efficient MATLAB computations with sparse and factored tensors. SIAM J Sci Comput, 2007, 30(1): 205-231

[4] Benetos E., Kotropoulos C.. Non-negative tensor factorization applied to music genre classification. IEEE Trans Audio, Speech, Language Processing, 2010, 18(8): 1955-1967

[5] Van Benthem M. H., Keenan M. R.. Fast algorithm for the solution of large-scale nonnegativity-constrained least squares problems. J Chemometrics, 2004, 18(10): 441-450

[6] Berry M. W., Browne M.. Email surveillance using non-negative matrix factorization. Comput Math Organization Theory, 2005, 11(3): 249-264

[7] 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

[8] Boyd S, Parikh N, Chu E, Peleato B, Eckstein J. Distributed Optimization and Statistical Learning via the Alternating Direction Method of Multipliers. In: Jordan M, ed. Foundations and Trends in Machine Learning, Vol 3. 2011, 1–122 http://www.stanford.edu/~boyd/papers/admm_distr_stats.html

[9] Bro R., De Jong S.. A fast non-negativity-constrained least squares algorithm. J Chemometrics, 1997, 11(5): 393-401

[10] Chen Y., Wang X., Shi C., Lua E. K., Fu X. M., Deng B. X., Li X.. Phoenix: a weight-based network coordinate system using matrix factorization. IEEE Trans Network Service Management, 2011, 8(4): 334-347

[11] Cichocki A., Zdunek R., Phan A. H., Amari S.. Nonnegative Matrix and Tensor Factorizations: Applications to Exploratory Multi-way Data Analysis and Blind Source Separation, 2009, New York: Wiley

[12] Georghiades A. S., Belhumeur P. N., Kriegman D. J.. From few to many: illumination cone models for face recognition under variable lighting and pose. IEEE Trans Pattern Anal Machine Intelligence, 2001, 23(6): 643-660

[13] Han D. R., Xu W., Yang H.. An operator splitting method for variational inequalities with partially unknown mappings. Numer Math, 2008, 111(2): 207-237

[14] He B. S., Liao L. Z., Han D. R., Yang H.. A new inexact alternating directions method for monotone variational inequalities. Math Program, 2002, 92(1): 103-118

[15] He B. S., Yang H.. Some convergence properties of a method of multipliers for linearly constrained monotone variational inequalities. Oper Res Lett, 1998, 23(3–5): 151-161

[16] He B. S., Yang H., Wang S. L.. Alternating direction method with self-adaptive penalty parameters for monotone variational inequalities. J Optim Theory Appl, 2000, 106(2): 337-356

[17] Kim H., Park H.. Sparse non-negative matrix factorizations via alternating nonnegativity-constrained least squares for microarray data analysis. Bioinformatics, 2007, 23(12): 1495-1502

[18] Kim H., Park H.. Nonnegative matrix factorization based on alternating nonnegativity constrained least squares and active set method. SIAM J Matrix Anal Appl, 2008, 30(2): 713-730

[19] Kim J., Park H.. Fast nonnegative matrix factorization: an active-set-like method and comparisons. SIAM J Sci Comput, 2011, 33(6): 3261-3281

[20] Lawson C. L., Hanson R. J.. Solving Least Squares Problems, 1995, Philadelphia: SIAM

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

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

[23] Lee K.-C., Ho J., Kriegman D. J.. Acquiring linear subspaces for face recognition under variable lighting. IEEE Trans Pattern Anal Machine Intelligence, 2005, 27(5): 684-698

[24] Lim L. -H., Comon P.. Nonnegative approximations of nonnegative tensors. J Chemometrics, 2009, 23(7–8): 432-441

[25] Lin C. -J.. Projected gradient methods for non-negative matrix factorization. Neural Comput, 2007, 19(10): 2756-2779

[26] Mao Y., Saul L. K., Smith J. M.. IDES: an internet distance estimation service for large networks. IEEE J Selected Areas Communications, 2006, 24(12): 2273-2284

[27] Nielsen F. A., Balslev D., Hansen L. K.. Mining the posterior cingulate: segregation between memory and pain components. NeuroImage, 2005, 27(3): 520-532

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

[29] Schmidt M N, Mohamed S. Probabilistic non-negative tensor factorisation using markov chain monte carlo. In: European Signal Processing Conference. 2009, 1918–1922

[30] Shashua A, Hazan T. Non-negative tensor factorization with applications to statistics and computer vision. In: Proceedings of the 22nd International Conference on Machine Learning (ICML’ 05). 2005, 792–799

[31] Smilde A., Bro R., Geladi P.. Multi-way Analysis: Applications in the Chemical Sciences, 2004, New York: John Wiley & Sons

[32] Vavasis S. A.. On the complexity of nonnegative matrix factorization. SIAM J Optim, 2009, 20(3): 1364-1377

[33] Zhang Q., Wang H., Plemmons R., Pauca V. P.. Spectral unmixing using nonnegative tensor factorization. ACM Southeast Regional Conference, 2007, New York: ACM 531 532

[34] Zhang Y. Theory of compressive sensing via l₁-minimizatIon: a non-RIP analysis and extensions. Technical Report TR08-11, revised. Department of Computational and Applied Mathematics, Rice University, Houston, Texas. 2008. http://www.caam.rice.edu/~zhang/reports/tr0811_revised.pdf

[35] Zhang Y. An alternating direction algorithm for nonnegative matrix factorization. Technical Report TR10-03. Department of Computational and Applied Mathematics, Rice University, Houston, Texas. 2010. http://www.caam.rice.edu/?zhang/reports/tr1003.pdf