Optimal matrix-mimetic tensor algebras via variable projection

DOI10.1137/24M1702635MaRDI QIDQ6936082zbMATH OpenFDO

Authors Elizabeth Newman, Katherine Keegan

Publication date 2 September 2025

Published in SIAM Journal on Matrix Analysis and Applications (Search for Journal in Brave)

Full work available at URL https://doi.org/10.1137/24M1702635

zbMATH Keywords

linear transformationmachine learningmatrix-mimetic datamultilinear algebrareduced order modelingRiemannian optimizationtensor decompositionvariable projection

Mathematics Subject Classification ID

Vector and tensor algebra, theory of invariants (15A72) Numerical computation of eigenvalues and eigenvectors of matrices (65F15) Numerical linear algebra (65F99) Numerical optimization and variational techniques (65K10) Learning and adaptive systems in artificial intelligence (68T05)

Review text

This interesting paper studies optimal matrix-mimetic tensor algebras via variable projections. One motivation for doing this is that multilinear or tensor-based analogues to the singular value decomposition (SVD) are now known to be powerful tools to analyse multiway data. There are many reasons for this which are outlined in the paper under review. In the paper under review the authors produce a new algorithm to learn a linear transformation and desired tensor representation simultaneously. They call this new algorithm M-optimization which leverages matrix-mimetic optimality properties through variable projection, explicitly coupling the representation and the linear mapping. M-optimization does not rely on prior knowledge of the data. The authors also illustrate the generality their framework through numerical experiments on a wide range of applications, including financial index tracking, image compression, and reduced order modeling.\N\NThe paper is very well written with a very good set of references.

Cites work

• A Multilinear Singular Value Decomposition
• A randomized tensor singular value decomposition based on the t-product.
• A survey of projection-based model reduction methods for parametric dynamical systems
• A variable projection method for solving separable nonlinear least squares problems
• An efficient computational approach for multiframe blind deconvolution
• An Introduction to Optimization on Smooth Manifolds
• Analysis of individual differences in multidimensional scaling via an \(n\)-way generalization of ``Eckart-Young decomposition
• Approximate joint diagonalization with Riemannian optimization on the general linear group
• Factorization strategies for third-order tensors
• Functions of Matrices
• Global rates of convergence for nonconvex optimization on manifolds
• Guaranteed minimum-rank solutions of linear matrix equations via nuclear norm minimization
• scientific article; zbMATH DE number 1049347 (Why is no real title available?)
• scientific article; zbMATH DE number 5223994 (Why is no real title available?)
• scientific article; zbMATH DE number 961607 (Why is no real title available?)
• scientific article; zbMATH DE number 3067835 (Why is no real title available?)
• Introduction to applied linear algebra. Vectors, matrices, and least squares
• Local improvements to reduced-order models using sensitivity analysis of the proper orthogonal decomposition
• Nonnegative tensor patch dictionary approaches for image compression and deblurring applications
• On the Perturbation of Pseudo-Inverses, Projections and Linear Least Squares Problems
• Randomized Kaczmarz for tensor linear systems
• Reduced order modeling based shape optimization of surface acoustic wave driven microfluidic biochips
• Robust tensor completion using transformed tensor singular value decomposition
• Sparse Portfolios for High-Dimensional Financial Index Tracking
• Tensor Decompositions and Applications
• Tensor Q-rank: new data dependent definition of tensor rank
• Tensor-tensor algebra for optimal representation and compression of multiway data
• Tensor-tensor products with invertible linear transforms
• Tensor-train decomposition
• The approximation of one matrix by another of lower rank.
• The Differentiation of Pseudo-Inverses and Nonlinear Least Squares Problems Whose Variables Separate
• The expression of a tensor or a polyadic as a sum of products.
• The Geometry of Algorithms with Orthogonality Constraints
• The power and Arnoldi methods in an algebra of circulants.
• Third-order tensors as operators on matrices: a theoretical and computational framework with applications in imaging
• Train Like a (Var)Pro: Efficient Training of Neural Networks with Variable Projection
• Turbulence and the dynamics of coherent structures. I. Coherent structures
• Variable projection for nonlinear least squares problems
• Variable Projection for NonSmooth Problems

Cited in

(1)

• Robust Low-Tubal-Rank Tensor Completion under Cross-Concentrated Sampling

This page was built for publication: Optimal matrix-mimetic tensor algebras via variable projection

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6936082)