Higher-order singular value decomposition: Difference between revisions

Revision as of 01:56, 16 June 2025

Revision as of 21:16, 16 June 2025

Line 63:

}}</ref><ref name=":2" /> are sequential algorithms that use gradient descent and the power method, respectively.

=== M-mode SVD (oftenalso referred toknown as HOSVD or Tucker)===

What is commonly referred to as the HOSVD or Tucker was developed by [[M.A.O. Vasilescu]] and [[Demetri Terzopoulos|Terzopoulos]] under

the name M-mode SVD<ref name=":Vasilescu2002"/><ref name=":Vasilescu2005"/>. <br/>The

M-mode SVD is a simple elegant algorithm suitable for parallel computation, but often. referredHowever, intoit theis literaturefrequently asconflated with the TuckerHOSVD or the HOSVDTucker decompositions, which are sequentialtypically algorithmsimplemented thatas employsequential gradientalgorithms descentrelying oron the power method or gradient descent, respectively.

* For <math>m=1,\ldots,M</math>, do the following:

# Construct the mode-''m'' flattening <math>\mathcal{A}_{[m]}</math>;