Big Data Matrix Singular Value Decomposition Based on Low-Rank Tensor Train Decomposition

被引:5
作者
Lee, Namgil [1 ]
Cichocki, Andrzej [1 ]
机构
[1] RIKEN, Brain Sci Inst, Lab Adv Brain Signal Proc, Wako, Saitama 3510198, Japan
来源
ADVANCES IN NEURAL NETWORKS - ISNN 2014 | 2014年 / 8866卷
关键词
Big data processing; Curse-of-dimensionality; Eigenvalue decomposition; Singular value decomposition; Optimization; Tensor network; Matrix product states; VECTORS;
D O I
10.1007/978-3-319-12436-0_14
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
We propose singular value decomposition (SVD) algorithms for very large-scale matrices based on a low-rank tensor decomposition technique called the tensor train (TT) format. By using the proposed algorithms, we can compute several dominant singular values and corresponding singular vectors of large-scale structured matrices given in a low-rank TT format. We propose a large-scale trace optimization problem, and in the proposed methods, the large-scale optimization problem is reduced to sequential small-scale optimization problems. We show that the computational complexity of the proposed algorithms scales logarithmically with the matrix size if the TT-ranks are bounded. Numerical simulations based on very large-scale Hilbert matrix demonstrate the effectiveness of the proposed methods.
引用
收藏
页码:121 / 130
页数:10
相关论文
共 17 条
[1]  
[Anonymous], 2013, LIT SURVEY LOWRANK T
[2]  
[Anonymous], 1998, Solution of large-scale eigenvalue problems with implicitly restarted Arnoldi methods, DOI DOI 10.1137/1.9780898719628
[3]  
Cichocki A., 2014, ARXIV13016068
[4]   TRACKING A FEW EXTREME SINGULAR-VALUES AND VECTORS IN SIGNAL-PROCESSING [J].
COMON, P ;
GOLUB, GH .
PROCEEDINGS OF THE IEEE, 1990, 78 (08) :1327-1343
[5]   Computation of extreme eigenvalues in higher dimensions using block tensor train format [J].
Dolgov, S. V. ;
Khoromskij, B. N. ;
Oseledets, I. V. ;
Savostyanov, D. V. .
COMPUTER PHYSICS COMMUNICATIONS, 2014, 185 (04) :1207-1216
[6]   Fast Monte-Carlo algorithms for finding low-rank approximations [J].
Frieze, A ;
Kannan, R ;
Vempala, S .
39TH ANNUAL SYMPOSIUM ON FOUNDATIONS OF COMPUTER SCIENCE, PROCEEDINGS, 1998, :370-378
[7]   THE ALTERNATING LINEAR SCHEME FOR TENSOR OPTIMIZATION IN THE TENSOR TRAIN FORMAT [J].
Holtz, Sebastian ;
Rohwedder, Thorsten ;
Schneider, Reinhold .
SIAM JOURNAL ON SCIENTIFIC COMPUTING, 2012, 34 (02) :A683-A713
[8]   On manifolds of tensors of fixed TT-rank [J].
Holtz, Sebastian ;
Rohwedder, Thorsten ;
Schneider, Reinhold .
NUMERISCHE MATHEMATIK, 2012, 120 (04) :701-731
[9]   Computations in quantum tensor networks [J].
Huckle, T. ;
Waldherr, K. ;
Schulte-Herbrueggen, T. .
LINEAR ALGEBRA AND ITS APPLICATIONS, 2013, 438 (02) :750-781
[10]   MULTILEVEL TOEPLITZ MATRICES GENERATED BY TENSOR-STRUCTURED VECTORS AND CONVOLUTION WITH LOGARITHMIC COMPLEXITY [J].
Kazeev, Vladimir A. ;
Khoromskij, Boris N. ;
Tyrtyshnikov, Eugene E. .
SIAM JOURNAL ON SCIENTIFIC COMPUTING, 2013, 35 (03) :A1511-A1536