Canonical Polyadic Decomposition with a Columnwise Orthonormal Factor Matrix

@article{Srensen2012CanonicalPD,
  title={Canonical Polyadic Decomposition with a Columnwise Orthonormal Factor Matrix},
  author={Mikael S{\o}rensen and Lieven De Lathauwer and Pierre Comon and Sylvie Icart and Luc Deneire},
  journal={SIAM J. Matrix Anal. Appl.},
  year={2012},
  volume={33},
  pages={1190-1213}
}
Canonical polyadic decomposition (CPD) of a higher-order tensor is an important tool in mathematical engineering. In many applications at least one of the matrix factors is constrained to be columnwise orthonormal. We first derive a relaxed condition that guarantees uniqueness of the CPD under this constraint. Second, we give a simple proof of the existence of the optimal low-rank approximation of a tensor in the case that a factor matrix is columnwise orthonormal. Third, we derive numerical… 

Figures and Tables from this paper

2 IGNAT DOMANOV AND

This paper presents an algebraic algorithm for the computation of the CPD in cases where none of the factor matrices has full column rank and shows that if the famous Kruskal condition holds, then the C PD can be found algebraically.

Blind Signal Separation via Tensor Decomposition With Vandermonde Factor: Canonical Polyadic Decomposition

It is explained that, under new, relaxed uniqueness conditions, the number of components may simply be estimated as the rank of a matrix and an efficient algorithm for the computation of the factors that only resorts to basic linear algebra is proposed.

Rank Properties and Computational Methods for Orthogonal Tensor Decompositions

  • Chao Zeng
  • Computer Science, Mathematics
    Journal of Scientific Computing
  • 2022
This work presents several properties of orthogonal rank, which are different from those of tensor rank in many aspects, and proposes an algorithm based on the augmented Lagrangian method that has a great advantage over the existing methods for strongly Orthogonal decompositions in terms of the approximation error.

Study of Canonical Polyadic Decomposition of Higher-Order Tensors (Studie van de canonieke polyadische decompositie van hogere-orde tensoren)

The uniqueness of structured matrix factorizations (and hence, the uniqueness of decompositions of a matrix into a sum of structured rank-1 matrices) has found many applications in engineering and is well known.

Cramér-Rao-Induced Bounds for CANDECOMP/PARAFAC Tensor Decomposition

A Cramér-Rao lower bound (CRLB) on the variance of unbiased estimates of factor matrices in Canonical Polyadic (CP) or CANDECOMP/PARAFAC decompositions of a tensor from noisy observations is presented.

Probabilistic Tensor Canonical Polyadic Decomposition With Orthogonal Factors

A novel tensor CPD algorithm based on the probabilistic inference framework is devised, based on which an inference algorithm is proposed that alternatively estimates the factor matrices, recovers the tensor rank, and mitigates the outliers.

Guarantees for existence of a best canonical polyadic approximation of a noisy low-rank tensor

Deterministic bounds for the existence of best low rank tensor approximations over K = R or K = C are given and it is shown that every K-rank R tensor inside of this ball has a unique canonical polyadic decomposition.

Rank-one tensor injection: A novel method for canonical polyadic tensor decomposition

A method to automatically generate the injected tensor which satisfies a specific estimation accuracy such that this tensor should not dominate rank-1 tensors of the data tensor, but is still able to be retrieved with a sufficient accuracy.

Half-Quadratic Alternating Direction Method of Multipliers for Robust Orthogonal Tensor Approximation

This paper derives a robust orthogonal tensor CPD model with Cauchy loss, which is resistant to heavy-tailed noise or outliers and shows that the whole sequence generated by the algorithm globally converges to a stationary point of the problem under consideration.

Linear convergence of an alternating polar decomposition method for low rank orthogonal tensor approximations

An improved version iAPD of the classical APD is proposed, which exhibits an overall sublinear convergence with an explicit rate which is sharper than the usual $O(1/k)$ for first order methods in optimization.
...

References

SHOWING 1-10 OF 42 REFERENCES

A Link between the Canonical Decomposition in Multilinear Algebra and Simultaneous Matrix Diagonalization

This paper derives a new and relatively weak deterministic sufficient condition for uniqueness in the decomposition of higher-order tensors which have the property that the rank is smaller than the greatest dimension.

On the Tensor SVD and the Optimal Low Rank Orthogonal Approximation of Tensors

The existence of an optimal approximation is theoretically guaranteed under certain conditions, and this optimal approximation yields a tensor decomposition where the diagonal of the core is maximized.

On the Best Rank-1 and Rank-(R1 , R2, ... , RN) Approximation of Higher-Order Tensors

A multilinear generalization of the best rank-R approximation problem for matrices, namely, the approximation of a given higher-order tensor, in an optimal least-squares sense, by a tensor that has prespecified column rank value, rowRank value, etc.

A Multilinear 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.

Canonical decomposition of even higher order cumulant arrays for blind underdetermined mixture identification

A new family of methods, named 2q-ORBIT (q > 1), is proposed in this paper in order to blindly identify potentially underdetermined mixtures of statistically independent sources. These methods are

Kruskal's permutation lemma and the identification of CANDECOMP/PARAFAC and bilinear models with constant modulus constraints

Kruskal's permutation lemma is revisited, and a similar necessary and sufficient condition for unique bilinear factorization under constant modulus (CM) constraints is derived, thus providing an interesting link to (and unification with) CP.

On Uniqueness of the nth Order Tensor Decomposition into Rank-1 Terms with Linear Independence in One Mode

  • A. Stegeman
  • Mathematics, Computer Science
    SIAM J. Matrix Anal. Appl.
  • 2010
The results include necessary, sufficient, necessary and sufficient, and generic uniqueness conditions, which have the form of a bound on $R$ in terms of the dimensions of the tensor to be decomposed.

Quasi-Newton Methods on Grassmannians and Multilinear Approximations of Tensors

BFGS and limited memory BFGS updates in local and global coordinates on Grassmannians or a product of these are defined and it is proved that, when local coordinates are used, their BF GS updates on Grassmanians share the same optimality property as the usual BFGS Updates on Euclidean spaces.

Tucker compression and local optima