JMLR

A Random Matrix Approach to Low-Multilinear-Rank Tensor Approximation

Authors
Hugo Lebeau Florent Chatelain Romain Couillet
Paper Information
  • Journal:
    Journal of Machine Learning Research
  • Added to Tracker:
    Jul 30, 2025
Abstract

This work presents a comprehensive understanding of the estimation of a planted low-rank signal from a general spiked tensor model near the computational threshold. Relying on standard tools from the theory of large random matrices, we characterize the large-dimensional spectral behavior of the unfoldings of the data tensor and exhibit relevant signal-to-noise ratios governing the detectability of the principal directions of the signal. These results allow to accurately predict the reconstruction performance of truncated multilinear SVD (MLSVD) in the non-trivial regime. This is particularly important since it serves as an initialization of the higher-order orthogonal iteration (HOOI) scheme, whose convergence to the best low-multilinear-rank approximation depends entirely on its initialization. We give a sufficient condition for the convergence of HOOI and show that the number of iterations before convergence tends to $1$ in the large-dimensional limit.

Author Details
Hugo Lebeau
Author
Florent Chatelain
Author
Romain Couillet
Author
Citation Information
APA Format
Hugo Lebeau , Florent Chatelain & Romain Couillet . A Random Matrix Approach to Low-Multilinear-Rank Tensor Approximation. Journal of Machine Learning Research .
BibTeX Format
@article{paper313,
  title = { A Random Matrix Approach to Low-Multilinear-Rank Tensor Approximation },
  author = { Hugo Lebeau and Florent Chatelain and Romain Couillet },
  journal = { Journal of Machine Learning Research },
  url = { https://www.jmlr.org/papers/v26/24-0193.html }
}