Robust principal component analysis?

Journal of the ACM - Tập 58 Số 3 - Trang 1-37 - 2011
Emmanuel J. Candès1, Xiaodong Li1, Yi Ma2, John Wright3
1Stanford University, Stanford, CA
2University of Illinois at Urbana-Champaign, Urbana, IL, Microsoft Research Asia, Beijing, China#TAB#
3Microsoft Research Asia, Beijing, China

Tóm tắt

This article is about a curious phenomenon. Suppose we have a data matrix, which is the superposition of a low-rank component and a sparse component. Can we recover each component individually? We prove that under some suitable assumptions, it is possible to recover both the low-rank and the sparse components exactly by solving a very convenient convex program called Principal Component Pursuit ; among all feasible decompositions, simply minimize a weighted combination of the nuclear norm and of the ℓ 1 norm. This suggests the possibility of a principled approach to robust principal component analysis since our methodology and results assert that one can recover the principal components of a data matrix even though a positive fraction of its entries are arbitrarily corrupted. This extends to the situation where a fraction of the entries are missing as well. We discuss an algorithm for solving this optimization problem, and present applications in the area of video surveillance, where our methodology allows for the detection of objects in a cluttered background, and in the area of face recognition, where it offers a principled way of removing shadows and specularities in images of faces.

Từ khóa


Tài liệu tham khảo

10.1109/TPAMI.2003.1177153

10.1137/080716542

10.1137/090756855

10.1162/089976603321780317

Bertsekas , D. 1982. Constrained Optimization and Lagrange Multiplier Method . Academic Press . Bertsekas, D. 1982. Constrained Optimization and Lagrange Multiplier Method. Academic Press.

10.1137/080738970

10.1109/JPROC.2009.2035722

10.1007/s10208-009-9045-5

10.1109/TIT.2005.862083

10.1109/TIT.2010.2044061

10.1007/978-3-540-88688-4_12

Chandrasekaran V. Sanghavi S. Parrilo P. and Willsky A. 2009. Rank-sparsity incoherence for matrix decomposition. Siam J. Optim. to appear http://arxiv.org/abs/0906.2220. Chandrasekaran V. Sanghavi S. Parrilo P. and Willsky A. 2009. Rank-sparsity incoherence for matrix decomposition. Siam J. Optim. to appear http://arxiv.org/abs/0906.2220.

10.1137/S003614450037906X

10.1002/(SICI)1097-4571(199009)41:6<391::AID-ASI1>3.0.CO;2-9

10.1007/BF02288367

Fazel , M. , Hindi , H. , and Boyd , S . 2003. Log-det heuristic for matrix rank minimization with applications to Hankel and Euclidean distance matrices . In Proceedings of the American Control Conference 2156--2162 . Fazel, M., Hindi, H., and Boyd, S. 2003. Log-det heuristic for matrix rank minimization with applications to Hankel and Euclidean distance matrices. In Proceedings of the American Control Conference 2156--2162.

10.1145/358669.358692

10.1109/34.927464

10.2307/2528963

Goldfarb D. and Ma S. 2009. Convergence of fixed point continuation algorithms for matrix rank minimization. http://arxiv.org/abs/0906.3499. Goldfarb D. and Ma S. 2009. Convergence of fixed point continuation algorithms for matrix rank minimization. http://arxiv.org/abs/0906.3499.

Grant , M. , and Boyd , S . 2009 . CVX: Matlab software for disciplined convex programming (web page and software) . http://stanford.edu/~boyd/cvx. Grant, M., and Boyd, S. 2009. CVX: Matlab software for disciplined convex programming (web page and software). http://stanford.edu/~boyd/cvx.

10.1109/TIT.2011.2104999

10.1103/PhysRevLett.105.150401

Hey T. Tansley S. and Tolle K. 2009. The Fourth Paradigm: Data-Intensive Scientific Discovery. Microsoft Research. Hey T. Tansley S. and Tolle K. 2009. The Fourth Paradigm: Data-Intensive Scientific Discovery. Microsoft Research.

10.1037/h0071325

Huber , P. 1981. Robust Statistics . Wiley . Huber, P. 1981. Robust Statistics. Wiley.

Jolliffe , I. 1986. Principal Component Analysis . Springer-Verlag . Jolliffe, I. 1986. Principal Component Analysis. Springer-Verlag.

10.1109/CVPR.2005.309

10.1109/TIT.2010.2046205

10.1007/BF02680549

Ledoux , M. 2001. The Concentration of Measure Phenomenon . American Mathematical Society . Ledoux, M. 2001. The Concentration of Measure Phenomenon. American Mathematical Society.

10.1109/TIP.2004.836169

Lin Z. Chen M. and Ma Y. 2009a. The augmented Lagrange multiplier method for exact recovery of a corrupted low-rank matrices. http://arxiv.org/abs/1009.5055. Lin Z. Chen M. and Ma Y. 2009a. The augmented Lagrange multiplier method for exact recovery of a corrupted low-rank matrices. http://arxiv.org/abs/1009.5055.

Lin , Z. , Ganesh , A. , Wright , J. , Wu , L. , Chen , M. , and Ma , Y . 2009b. Fast convex optimization algorithms for exact recovery of a corrupted low-rank matrix . In Proceedings of the Symposium on Computational Advances in Multi-Sensor Adaptive Processing (CAMSAP). Lin, Z., Ganesh, A., Wright, J., Wu, L., Chen, M., and Ma, Y. 2009b. Fast convex optimization algorithms for exact recovery of a corrupted low-rank matrix. In Proceedings of the Symposium on Computational Advances in Multi-Sensor Adaptive Processing (CAMSAP).

10.1137/0716071

10.1137/090755436

10.5555/3112668.3113011

Nesterov , Y. 1983 . A method of solving a convex programming problem with convergence rate O(1/k&lt;sup&gt;2&lt;/sup&gt;) . Soviet Math. Dokl. 27 , 2, 372 -- 376 . Nesterov, Y. 1983. A method of solving a convex programming problem with convergence rate O(1/k&lt;sup&gt;2&lt;/sup&gt;). Soviet Math. Dokl. 27, 2, 372--376.

10.1007/s10107-004-0552-5

Nesterov Y. 2007. Gradient methods for minimizing composite objective functions. Tech. rep. - CORE - Universite Catholique de Louvain. Nesterov Y. 2007. Gradient methods for minimizing composite objective functions. Tech. rep. - CORE - Universite Catholique de Louvain.

Netflix Inc. The Netflix prize. http://www.netflixprize.com/. Netflix Inc. The Netflix prize. http://www.netflixprize.com/.

10.1137/040605412

10.1006/jcss.2000.1711

Recht B. 2009. A simpler approach to matrix completion. CoRR abs/0910.0651. Recht B. 2009. A simpler approach to matrix completion. CoRR abs/0910.0651.

10.1137/070697835

Stauffer , C. , and Grimson , E . 1999. Adaptive background mixture models for real-time tracking . In Proceedings of the IEEE International Conference on Computer Vision and Pattern Recognition. Stauffer, C., and Grimson, E. 1999. Adaptive background mixture models for real-time tracking. In Proceedings of the IEEE International Conference on Computer Vision and Pattern Recognition.

Tenenbaum J. de Silva V. and Langford J. 2000. A global geometric framework for nonlinear dimensionality reduction. Science 290 5500 2319--2323. Tenenbaum J. de Silva V. and Langford J. 2000. A global geometric framework for nonlinear dimensionality reduction. Science 290 5500 2319--2323.

Toh , K. C. , and Yun , S. 2010 . An accelerated proximal gradient algorithm for nuclear norm regularized least squares problems . Pac. J. Optim. 6 , 615 -- 640 . Toh, K. C., and Yun, S. 2010. An accelerated proximal gradient algorithm for nuclear norm regularized least squares problems. Pac. J. Optim. 6, 615--640.

10.1023/A:1023709501986

Vershynin R. 2011. Introduction to the non-asymptotic analysis of random matrices. http://www-personal.umich.edu/&tilde;romanv/papers/non-asymptotic-rmt-plain.pdf. Vershynin R. 2011. Introduction to the non-asymptotic analysis of random matrices. http://www-personal.umich.edu/&tilde;romanv/papers/non-asymptotic-rmt-plain.pdf.

10.1137/070698920

10.1137/070703983

Yuan X. and Yang J. 2009. Sparse and low-rank matrix decomposition via alternating direction methods. http://www.optimization-online.org/08_HTML/2009/11/2447.html. Yuan X. and Yang J. 2009. Sparse and low-rank matrix decomposition via alternating direction methods. http://www.optimization-online.org/08_HTML/2009/11/2447.html.

Zhou , Z. , Wagner , A. , Mobahi , H. , Wright , J. , and Ma , Y . 2009. Face recognition with contiguous occlusion using Markov random fields . In Proceedings of the International Conference on Computer Vision (ICCV). Zhou, Z., Wagner, A., Mobahi, H., Wright, J., and Ma, Y. 2009. Face recognition with contiguous occlusion using Markov random fields. In Proceedings of the International Conference on Computer Vision (ICCV).

Thông tin
Thông tin xuất bản

Journal of the ACM

Tập 58 Số 3

1-37

Thông tin tác giả