Higher-order singular value decomposition

Summary

In multilinear algebra, the higher-order singular value decomposition (HOSVD) of a tensor is a specific orthogonal Tucker decomposition. It may be regarded as one type of generalization of the matrix singular value decomposition. It has applications in computer vision, computer graphics, machine learning, scientific computing, and signal processing. Some aspects can be traced as far back as F. L. Hitchcock in 1928,[1] but it was L. R. Tucker who developed for third-order tensors the general Tucker decomposition in the 1960s,[2][3][4] further advocated by L. De Lathauwer et al.[5] in their Multilinear SVD work that employs the power method, or advocated by Vasilescu and Terzopoulos that developed M-mode SVD a parallel algorithm that employs the matrix SVD.

The term higher order singular value decomposition (HOSVD) was coined be DeLathauwer, but the algorithm referred to commonly in the literature as the HOSVD and attributed to either Tucker or DeLathauwer was developed by Vasilescu and Terzopoulos.[6][7][8]Robust and L1-norm-based variants of HOSVD have also been proposed.[9][10][11][12]


Definition edit

For the purpose of this article, the abstract tensor   is assumed to be given in coordinates with respect to some basis as a M-way array, also denoted by  , where M is the number of modes and the order of the tensor.   is the complex numbers and it includes both the real numbers   and the pure imaginary numbers.

Let  be a unitary matrix containing a basis of the left singular vectors of the standard mode-m flattening   of   such that the jth column   of   corresponds to the jth largest singular value of  . Observe that the mode/factor matrix   does not depend on the particular on the specific definition of the mode m flattening. By the properties of the multilinear multiplication, we have

 
where   denotes the conjugate transpose. The second equality is because the  's are unitary matrices. Define now the core tensor
 
Then, the HOSVD[5] of   is the decomposition
 
The above construction shows that every tensor has a HOSVD.

Compact HOSVD edit

As in the case of the compact singular value decomposition of a matrix, it is also possible to consider a compact HOSVD, which is very useful in applications.

Assume that   is a matrix with unitary columns containing a basis of the left singular vectors corresponding to the nonzero singular values of the standard factor-m flattening   of  . Let the columns of   be sorted such that the   th column   of   corresponds to the  th largest nonzero singular value of  . Since the columns of   form a basis for the image of  , we have

 
where the first equality is due to the properties of orthogonal projections (in the Hermitian inner product) and the last equality is due to the properties of multilinear multiplication. As flattenings are bijective maps and the above formula is valid for all  , we find as before that
 
where the core tensor   is now of size  .

Multilinear rank edit

The multilinear rank[1] of   is denoted with rank- . The multilinear rank is a tuple in   where  . Not all tuples in   are multilinear ranks.[13] The multilinear ranks are bounded by   and it satisfy the constraint   must hold.[13]

The compact HOSVD is a rank-revealing deomposition in the sense that the dimensions of its core tensor correspond with the components of the multilinear rank of the tensor.

Interpretation edit

The following geometric interpretation is valid for both the full and compact HOSVD. Let   be the multilinear rank of the tensor  . Since   is a multidimensional array, we can expand it as follows

 
where   is the  th standard basis vector of  . By definition of the multilinear multiplication, it holds that
 
where the   are the columns of  . It is easy to verify that   is an orthonormal set of tensors. This means that the HOSVD can be interpreted as a way to express the tensor   with respect to a specifically chosen orthonormal basis   with the coefficients given as the multidimensional array  .

Computation edit

Let   be a tensor with a rank- , where   contains the reals   as a subset.

Classic computation edit

The strategy for computing the Multilinear SVD and the M-mode SVD was introduced in the 1960s by L. R. Tucker,[3] further advocated by L. De Lathauwer et al.,[5] and by Vasilescu and Terzopulous.[8][6] The term HOSVD was coined by Lieven De Lathauwer, but the algorithm typically referred to in the literature as HOSVD was introduced by Vasilescu and Terzopoulos[6][8] with the name M-mode SVD. It is a parallel computation that employs the matrix SVD to compute the orthonormal mode matrices.

M-mode SVD edit

Sources:[6][8]

  • For  , do the following:
  1. Construct the mode-m flattening  ;
  2. Compute the (compact) singular value decomposition  , and store the left singular vectors  ;
  • Compute the core tensor   via the multilinear multiplication  

Interlacing computation edit

A strategy that is significantly faster when some or all   consists of interlacing the computation of the core tensor and the factor matrices, as follows:[14][15][16]

  • Set  ;
  • For   perform the following:
    1. Construct the standard mode-m flattening  ;
    2. Compute the (compact) singular value decomposition  , and store the left singular vectors  ;
    3. Set  , or, equivalently,  .

In-place computation edit

The HOSVD can be computed in-place via the Fused In-place Sequentially Truncated Higher Order Singular Value Decomposition (FIST-HOSVD) [16] algorithm by overwriting the original tensor by the HOSVD core tensor, significantly reducing the memory consumption of computing HOSVD.

Approximation edit

In applications, such as those mentioned below, a common problem consists of approximating a given tensor   by one with a reduced multilinear rank. Formally, if the multilinear rank of   is denoted by  , then computing the optimal   that approximates   for a given reduced   is a nonlinear non-convex  -optimization problem

 
where   is the reduced multilinear rank with  , and the norm   is the Frobenius norm.

A simple idea for trying to solve this optimization problem is to truncate the (compact) SVD in step 2 of either the classic or the interlaced computation. A classically truncated HOSVD is obtained by replacing step 2 in the classic computation by

  • Compute a rank-  truncated SVD  , and store the top   left singular vectors  ;

while a sequentially truncated HOSVD (or successively truncated HOSVD) is obtained by replacing step 2 in the interlaced computation by

  • Compute a rank-  truncated SVD  , and store the top   left singular vectors  . Unfortunately, truncation does not result in an optimal solution for the best low multilinear rank optimization problem,[5][6][14][16]. However, both the classically and interleaved truncated HOSVD result in a quasi-optimal solution:[14][16][7][15][17] if   denotes the classically or sequentially truncated HOSVD and   denotes the optimal solution to the best low multilinear rank approximation problem, then
     
    in practice this means that if there exists an optimal solution with a small error, then a truncated HOSVD will for many intended purposes also yield a sufficiently good solution.

Applications edit

The HOSVD is most commonly applied to the extraction of relevant information from multi-way arrays.

Starting in the early 2000s, Vasilescu addressed causal questions by reframing the data analysis, recognition and synthesis problems as multilinear tensor problems. The power of the tensor framework was showcased by decomposing and representing an image in terms of its causal factors of data formation, in the context of Human Motion Signatures for gait recognition,[18] face recognition—TensorFaces[19][20] and computer graphics—TensorTextures.[21]

The HOSVD has been successfully applied to signal processing and big data, e.g., in genomic signal processing.[22][23][24] These applications also inspired a higher-order GSVD (HO GSVD)[25] and a tensor GSVD.[26]

A combination of HOSVD and SVD also has been applied for real-time event detection from complex data streams (multivariate data with space and time dimensions) in disease surveillance.[27]

It is also used in tensor product model transformation-based controller design.[28][29]

The concept of HOSVD was carried over to functions by Baranyi and Yam via the TP model transformation.[28][29] This extension led to the definition of the HOSVD-based canonical form of tensor product functions and Linear Parameter Varying system models[30] and to convex hull manipulation based control optimization theory, see TP model transformation in control theories.

HOSVD was proposed to be applied to multi-view data analysis[31] and was successfully applied to in silico drug discovery from gene expression.[32]

Robust L1-norm variant edit

L1-Tucker is the L1-norm-based, robust variant of Tucker decomposition.[10][11] L1-HOSVD is the analogous of HOSVD for the solution to L1-Tucker.[10][12]

References edit

  1. ^ a b Hitchcock, Frank L (1928-04-01). "Multiple Invariants and Generalized Rank of a M-Way Array or Tensor". Journal of Mathematics and Physics. 7 (1–4): 39–79. doi:10.1002/sapm19287139. ISSN 1467-9590.
  2. ^ Tucker, Ledyard R. (1966-09-01). "Some mathematical notes on three-mode factor analysis". Psychometrika. 31 (3): 279–311. doi:10.1007/bf02289464. ISSN 0033-3123. PMID 5221127. S2CID 44301099.
  3. ^ a b Tucker, L. R. (1963). "Implications of factor analysis of three-way matrices for measurement of change". In C. W. Harris (Ed.), Problems in Measuring Change. Madison, Wis.: Univ. Wis. Press.: 122–137.
  4. ^ Tucker, L. R. (1964). "The extension of factor analysis to three-dimensional matrices". In N. Frederiksen and H. Gulliksen (Eds.), Contributions to Mathematical Psychology. New York: Holt, Rinehart and Winston: 109–127.
  5. ^ a b c d De Lathauwer, L.; De Moor, B.; Vandewalle, J. (2000-01-01). "A Multilinear Singular Value Decomposition". SIAM Journal on Matrix Analysis and Applications. 21 (4): 1253–1278. CiteSeerX 10.1.1.102.9135. doi:10.1137/s0895479896305696. ISSN 0895-4798.
  6. ^ a b c d e M. A. O. Vasilescu, D. Terzopoulos (2002) with the name M-mode SVD. The M-mode SVD is suitable for parallel computation and employs the matrix SVD "Multilinear Analysis of Image Ensembles: TensorFaces", Proc. 7th European Conference on Computer Vision (ECCV'02), Copenhagen, Denmark, May, 2002
  7. ^ a b M. A. O. Vasilescu, D. Terzopoulos (2003) "Multilinear Subspace Analysis of Image Ensembles", "Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR’03), Madison, WI, June, 2003"
  8. ^ a b c d M. A. O. Vasilescu, D. Terzopoulos (2005) "Multilinear Independent Component Analysis", "Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR’05), San Diego, CA, June 2005, vol.1, 547–553."
  9. ^ Godfarb, Donald; Zhiwei, Qin (2014). "Robust low-rank tensor recovery: Models and algorithms". SIAM Journal on Matrix Analysis and Applications. 35 (1): 225–253. arXiv:1311.6182. doi:10.1137/130905010. S2CID 1051205.
  10. ^ a b c Chachlakis, Dimitris G.; Prater-Bennette, Ashley; Markopoulos, Panos P. (22 November 2019). "L1-norm Tucker Tensor Decomposition". IEEE Access. 7: 178454–178465. arXiv:1904.06455. doi:10.1109/ACCESS.2019.2955134.
  11. ^ a b Markopoulos, Panos P.; Chachlakis, Dimitris G.; Papalexakis, Evangelos (April 2018). "The Exact Solution to Rank-1 L1-Norm TUCKER2 Decomposition". IEEE Signal Processing Letters. 25 (4): 511–515. arXiv:1710.11306. Bibcode:2018ISPL...25..511M. doi:10.1109/LSP.2018.2790901. S2CID 3693326.
  12. ^ a b Markopoulos, Panos P.; Chachlakis, Dimitris G.; Prater-Bennette, Ashley (21 February 2019). "L1-Norm Higher-Order Singular-Value Decomposition". 2018 IEEE Global Conference on Signal and Information Processing (GlobalSIP). pp. 1353–1357. doi:10.1109/GlobalSIP.2018.8646385. ISBN 978-1-7281-1295-4. S2CID 67874182.
  13. ^ a b Carlini, Enrico; Kleppe, Johannes (2011). "Ranks derived from multilinear maps". Journal of Pure and Applied Algebra. 215 (8): 1999–2004. doi:10.1016/j.jpaa.2010.11.010.
  14. ^ a b c Vannieuwenhoven, N.; Vandebril, R.; Meerbergen, K. (2012-01-01). "A New Truncation Strategy for the Higher-Order Singular Value Decomposition". SIAM Journal on Scientific Computing. 34 (2): A1027–A1052. Bibcode:2012SJSC...34A1027V. doi:10.1137/110836067. ISSN 1064-8275. S2CID 15318433.
  15. ^ a b Hackbusch, Wolfgang (2012). Tensor Spaces and Numerical Tensor Calculus | SpringerLink. Springer Series in Computational Mathematics. Vol. 42. doi:10.1007/978-3-642-28027-6. ISBN 978-3-642-28026-9. S2CID 117253621.
  16. ^ a b c d Cobb, Benjamin; Kolla, Hemanth; Phipps, Eric; Çatalyürek, Ümit V. (2022). FIST-HOSVD: Fused in-Place Sequentially Truncated Higher Order Singular Value Decomposition. Platform for Advanced Scientific Computing(PASC). doi:10.1145/3539781.3539798. ISBN 9781450394109.
  17. ^ Grasedyck, L. (2010-01-01). "Hierarchical Singular Value Decomposition of Tensors". SIAM Journal on Matrix Analysis and Applications. 31 (4): 2029–2054. CiteSeerX 10.1.1.660.8333. doi:10.1137/090764189. ISSN 0895-4798.
  18. ^ M. A. O. Vasilescu (2002) "Human Motion Signatures: Analysis, Synthesis, Recognition," Proceedings of International Conference on Pattern Recognition (ICPR 2002), Vol. 3, Quebec City, Canada, Aug, 2002, 456–460.
  19. ^ M.A.O. Vasilescu, D. Terzopoulos (2003) "Multilinear Subspace Analysis for Image Ensembles, M. A. O. Vasilescu, D. Terzopoulos, Proc. Computer Vision and Pattern Recognition Conf. (CVPR '03), Vol.2, Madison, WI, June, 2003, 93–99.
  20. ^ M.A.O. Vasilescu, D. Terzopoulos (2002) "Multilinear Analysis of Image Ensembles: TensorFaces," Proc. 7th European Conference on Computer Vision (ECCV'02), Copenhagen, Denmark, May, 2002, in Computer Vision -- ECCV 2002, Lecture Notes in Computer Science, Vol. 2350, A. Heyden et al. (Eds.), Springer-Verlag, Berlin, 2002, 447–460.
  21. ^ M.A.O. Vasilescu, D. Terzopoulos (2004) "TensorTextures: Multilinear Image-Based Rendering", M. A. O. Vasilescu and D. Terzopoulos, Proc. ACM SIGGRAPH 2004 Conference Los Angeles, CA, August, 2004, in Computer Graphics Proceedings, Annual Conference Series, 2004, 336–342.
  22. ^ L. Omberg; G. H. Golub; O. Alter (November 2007). "A Tensor Higher-Order Singular Value Decomposition for Integrative Analysis of DNA Microarray Data From Different Studies". PNAS. 104 (47): 18371–18376. Bibcode:2007PNAS..10418371O. doi:10.1073/pnas.0709146104. PMC 2147680. PMID 18003902.
  23. ^ L. Omberg; J. R. Meyerson; K. Kobayashi; L. S. Drury; J. F. X. Diffley; O. Alter (October 2009). "Global Effects of DNA Replication and DNA Replication Origin Activity on Eukaryotic Gene Expression". Molecular Systems Biology. 5: 312. doi:10.1038/msb.2009.70. PMC 2779084. PMID 19888207. Highlight.
  24. ^ C. Muralidhara; A. M. Gross; R. R. Gutell; O. Alter (April 2011). "Tensor Decomposition Reveals Concurrent Evolutionary Convergences and Divergences and Correlations with Structural Motifs in Ribosomal RNA". PLOS ONE. 6 (4): e18768. Bibcode:2011PLoSO...618768M. doi:10.1371/journal.pone.0018768. PMC 3094155. PMID 21625625. Highlight.
  25. ^ S. P. Ponnapalli; M. A. Saunders; C. F. Van Loan; O. Alter (December 2011). "A Higher-Order Generalized Singular Value Decomposition for Comparison of Global mRNA Expression from Multiple Organisms". PLOS ONE. 6 (12): e28072. Bibcode:2011PLoSO...628072P. doi:10.1371/journal.pone.0028072. PMC 3245232. PMID 22216090. Highlight.
  26. ^ P. Sankaranarayanan; T. E. Schomay; K. A. Aiello; O. Alter (April 2015). "Tensor GSVD of Patient- and Platform-Matched Tumor and Normal DNA Copy-Number Profiles Uncovers Chromosome Arm-Wide Patterns of Tumor-Exclusive Platform-Consistent Alterations Encoding for Cell Transformation and Predicting Ovarian Cancer Survival". PLOS ONE. 10 (4): e0121396. Bibcode:2015PLoSO..1021396S. doi:10.1371/journal.pone.0121396. PMC 4398562. PMID 25875127. AAAS EurekAlert! Press Release and NAE Podcast Feature.
  27. ^ Hadi Fanaee-T; João Gama (May 2015). "EigenEvent: An algorithm for event detection from complex data streams in Syndromic surveillance". Intelligent Data Analysis. 19 (3): 597–616. arXiv:1406.3496. Bibcode:2014arXiv1406.3496F. doi:10.3233/IDA-150734. S2CID 17966555.
  28. ^ a b P. Baranyi (April 2004). "TP model transformation as a way to LMI based controller design". IEEE Transactions on Industrial Electronics. 51 (2): 387–400. doi:10.1109/tie.2003.822037. S2CID 7957799.
  29. ^ a b P. Baranyi; D. Tikk; Y. Yam; R. J. Patton (2003). "From Differential Equations to PDC Controller Design via Numerical Transformation". Computers in Industry. 51 (3): 281–297. doi:10.1016/s0166-3615(03)00058-7.
  30. ^ P. Baranyi; L. Szeidl; P. Várlaki; Y. Yam (July 3–5, 2006). Definition of the HOSVD-based canonical form of polytopic dynamic models. 3rd International Conference on Mechatronics (ICM 2006). Budapest, Hungary. pp. 660–665.
  31. ^ Y-h. Taguchi (August 2017). "Tensor decomposition-based unsupervised feature extraction applied to matrix products for multi-view data processing". PLOS ONE. 12 (8): e0183933. Bibcode:2017PLoSO..1283933T. doi:10.1371/journal.pone.0183933. PMC 5571984. PMID 28841719.
  32. ^ Y-h. Taguchi (October 2017). "Identification of candidate drugs using tensor-decomposition-based unsupervised feature extraction in integrated analysis of gene expression between diseases and DrugMatrix datasets". Scientific Reports. 7 (1): 13733. Bibcode:2017NatSR...713733T. doi:10.1038/s41598-017-13003-0. PMC 5653784. PMID 29062063.