From multilinear SVD to multilinear UTV decomposition
نویسندگان
چکیده
Across a range of applications, low multilinear rank approximation (LMLRA) is used to compress large tensors into more compact form, while preserving most their information. A specific instance LMLRA the singular value decomposition (MLSVD), which can be for principal component analysis (MLPCA). MLSVDs are obtained by computing SVDs all tensor unfoldings, but, in practical it often not necessary compute full SVDs. In this article, we therefore propose new decomposition, called truncated UTV (TMLUTVD). This that also rank-revealing, yet less expensive than MLSVD (TMLSVD); even computed finite number steps. We present its properties an algorithm-independent manner. particular, derive bounds on accuracy function truncation level. Experiments illustrate good performance practice.
منابع مشابه
A Multilinear Singular Value Decomposition
We discuss a multilinear generalization of the singular value decomposition. There is a strong analogy between several properties of the matrix and the higher-order tensor decomposition; uniqueness, link with the matrix eigenvalue decomposition, first-order perturbation effects, etc., are analyzed. We investigate how tensor symmetries affect the decomposition and propose a multilinear generaliz...
متن کاملMultilinear-NC1 6= Multilinear-NC2
An arithmetic circuit or formula is multilinear if the polynomial computed at each of its wires is multilinear. We give an explicit example for a polynomial f(x1, ..., xn), with coefficients in {0, 1}, such that over any field: 1. f can be computed by a polynomial-size multilinear circuit of depth O(log n). 2. Any multilinear formula for f is of size nΩ(logn). This gives a super-polynomial gap ...
متن کاملMultilinear Spectral Decomposition for Nonlinear Maxwell Equations
We study the frequency form, popular in physics, of Maxwell equations for nonlinear spatially inhomogeneous dielectric media. Using wavepackets based on eigenmodes of the underlying linear dielectric medium we develop a mathematically consistent interpretation of the frequency form of nonlinear Maxwell operators. In particular, we construct the operator of nonlinear polarization based on freque...
متن کاملShift-invariant multilinear decomposition of neuroimaging data
We present an algorithm for multilinear decomposition that allows for arbitrary shifts along one modality. The method is applied to neural activity arranged in the three modalities space, time, and trial. Thus, the algorithm models neural activity as a linear superposition of components with a fixed time course that may vary across either trials or space in its overall intensity and latency. It...
متن کاملOn the truncated multilinear singular value decomposition
In this report, we investigate the truncated multilinear singular value decomposition (MLSVD), proposed in De Lathauwer et al. (2000). Truncating the MLSVD results in an approximation, with a prescribed multilinear rank, to a tensor. We present a new error expression for an approximate Tucker decomposition with orthogonal factor matrices. From this expression, new insights are obtained which le...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: Signal Processing
سال: 2022
ISSN: ['0165-1684', '1872-7557']
DOI: https://doi.org/10.1016/j.sigpro.2022.108575