Multiresolution low-rank tensor formats
From MaRDI portal
Abstract: We describe a simple, black-box compression format for tensors with a multiscale structure. By representing the tensor as a sum of compressed tensors defined on increasingly coarse grids, we capture low-rank structures on each grid-scale, and we show how this leads to an increase in compression for a fixed accuracy. We devise an alternating algorithm to represent a given tensor in the multiresolution format and prove local convergence guarantees. In two dimensions, we provide examples that show that this approach can beat the Eckart-Young theorem, and for dimensions higher than two, we achieve higher compression than the tensor-train format on six real-world datasets. We also provide results on the closedness and stability of the tensor format and discuss how to perform common linear algebra operations on the level of the compressed tensors.
Recommendations
- Adaptive hierarchical subtensor partitioning for tensor compression
- Linear algebra for tensor problems
- Tensor train construction from tensor actions, with application to compression of large high order derivative tensors
- Random Projections for Low Multilinear Rank Tensors
- Low-rank tensor structure of linear diffusion operators in the TT and QTT formats
Cites work
- A Multilinear Singular Value Decomposition
- A Multiscale Data Assimilation with the Ensemble Kalman Filter
- A multiscale method for data assimilation
- Algorithm 862
- Efficient MATLAB Computations with Sparse and Factored Tensors
- scientific article; zbMATH DE number 5788282 (Why is no real title available?)
- scientific article; zbMATH DE number 6159604 (Why is no real title available?)
- Local convergence of the alternating least squares algorithm for canonical tensor approximation
- Multigrid accelerated tensor approximation of function related multidimensional arrays
- Multiscale methods for data assimilation in turbulent systems
- On manifolds of tensors of fixed TT-rank
- On the global convergence of the alternating least squares method for rank-one approximation to generic tensors
- On the Tensor SVD and the Optimal Low Rank Orthogonal Approximation of Tensors
- Orthogonal low rank tensor approximation: alternating least squares method and its global convergence
- Orthogonal tensor decompositions
- Randomized algorithms for the approximations of Tucker and the tensor train decompositions
- Rank-one approximation to high order tensors
- Riemannian optimization for high-dimensional tensor completion
- Stochastic superparameterization and multiscale filtering of turbulent tracers
- Tensor Decompositions and Applications
- Tensor decompositions for learning latent variable models
- Tensor numerical methods in scientific computing
- Tensor Rank and the Ill-Posedness of the Best Low-Rank Approximation Problem
- Tensor spaces and numerical tensor calculus
- Tensor-train decomposition
Cited in
(12)- A note on nonclosed tensor formats
- MERACLE: constructive layer-wise conversion of a tensor train into a MERA
- Low-rank tensor structure of linear diffusion operators in the TT and QTT formats
- Fast truncation of mode ranks for bilinear tensor operations
- Efficient construction of tensor ring representations from sampling
- Towards compact neural networks via end-to-end training: a Bayesian tensor approach with automatic rank determination
- Tensor CUR decomposition under T-product and its perturbation
- TuckerMPI: a parallel C++/MPI software package for large-scale data compression via the Tucker tensor decomposition
- Tensor train construction from tensor actions, with application to compression of large high order derivative tensors
- Adaptive hierarchical subtensor partitioning for tensor compression
- Algorithm 1036: ATC, an advanced Tucker compression library for multidimensional data
- Linear algebra for tensor problems
This page was built for publication: Multiresolution low-rank tensor formats
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5146611)
