Random Projections for Low Multilinear Rank Tensors

We propose two randomized tensor algorithms for reducing multilinear tensor rank. The basis of these randomized algorithms is from the work of Halko et al. (SIAM Rev 53(2):217–288, 2011). Here we provide some random versions of the higher order SVD and the higher order orthogonal iteration. Moreover, we provide a sharp probabilistic error bound for the matrix low rank approximation. In consequence, we provide an error bound for the tensor case. Moreover, we give several numerical examples which includes an implementation on a MRI dataset to test the efficacy of these randomized algorithms.

[1]  Berkant Savas,et al.  Quasi-Newton Methods on Grassmannians and Multilinear Approximations of Tensors , 2009, SIAM J. Sci. Comput..

[2]  Daniel Kressner,et al.  Low-Rank Tensor Krylov Subspace Methods for Parametrized Linear Systems , 2011, SIAM J. Matrix Anal. Appl..

[3]  Vin de Silva,et al.  Tensor rank and the ill-posedness of the best low-rank approximation problem , 2006, math/0607647.

[4]  Joos Vandewalle,et al.  A Multilinear Singular Value Decomposition , 2000, SIAM J. Matrix Anal. Appl..

[5]  Tamara G. Kolda,et al.  Tensor Decompositions and Applications , 2009, SIAM Rev..

[6]  Nathan Halko,et al.  Finding Structure with Randomness: Probabilistic Algorithms for Constructing Approximate Matrix Decompositions , 2009, SIAM Rev..

[7]  Christos Boutsidis,et al.  An improved approximation algorithm for the column subset selection problem , 2008, SODA.

[8]  Petros Drineas,et al.  Tensor-CUR Decompositions for Tensor-Based Data , 2008, SIAM J. Matrix Anal. Appl..

[9]  Tamara G. Kolda,et al.  Orthogonal Tensor Decompositions , 2000, SIAM J. Matrix Anal. Appl..

[10]  Christopher J. Hillar,et al.  Most Tensor Problems Are NP-Hard , 2009, JACM.

[11]  S. Geman.  A Limit Theorem for the Norm of Random Matrices , 1980 .

[12]  Na Li,et al.  Solving Multilinear Systems via Tensor Inversion , 2013, SIAM J. Matrix Anal. Appl..

[13]  O. Alter,et al.  A Higher-Order Generalized Singular Value Decomposition for Comparison of Global mRNA Expression from Multiple Organisms , 2011, PloS one.

[14]  P. Regalia.  Monotonically convergent algorithms for symmetric tensor approximation , 2013 .

[15]  C. Eckart,et al.  The approximation of one matrix by another of lower rank , 1936 .

[16]  Sarang C. Joshi,et al.  Detection of Crossing White Matter Fibers with High-Order Tensors and Rank-k Decompositions , 2011, IPMI.

[17]  Nadia Kreimer,et al.  A tensor higher-order singular value decomposition for prestack seismic data noise reduction and interpolation , 2012 .

[18]  Na Li,et al.  Source apportionment of time- and size-resolved ambient particulate matter , 2013 .

[19]  L. Tucker,et al.  Some mathematical notes on three-mode factor analysis , 1966, Psychometrika.

[20]  J. W. Silverstein.  The Smallest Eigenvalue of a Large Dimensional Wishart Matrix , 1985 .

[21]  Hans-Peter Seidel,et al.  Estimating Crossing Fibers: A Tensor Decomposition Approach , 2008, IEEE Transactions on Visualization and Computer Graphics.

[22]  Gene H. Golub,et al.  Matrix computations , 1983 .

[23]  Pierre Comon,et al.  Handbook of Blind Source Separation: Independent Component Analysis and Applications , 2010 .

[24]  M. Stephanov,et al.  Random Matrices , 2005, hep-ph/0509286.

[25]  Chun Chen,et al.  Hierarchical Tensor Approximation of Multi-Dimensional Visual Data , 2008, IEEE Transactions on Visualization and Computer Graphics.

[26]  Joos Vandewalle,et al.  On the Best Rank-1 and Rank-(R1 , R2, ... , RN) Approximation of Higher-Order Tensors , 2000, SIAM J. Matrix Anal. Appl..