Two-sided Matrix Regression - Archive ouverte HAL Accéder directement au contenu
Pré-Publication, Document De Travail Année : 2023

Two-sided Matrix Regression

Résumé

The two-sided matrix regression model Y = A * XB * + E aims at predicting Y by taking into account both linear links between column features of X, via the unknown matrix B * , and also among the row features of X, via the matrix A *. We propose lowrank predictors in this high-dimensional matrix regression model via rank-penalized and nuclear norm-penalized least squares. Both criteria are non jointly convex; however, we propose explicit predictors based on SVD and show optimal prediction bounds. We give sufficient conditions for consistent rank selector. We also propose a fully data-driven rank-adaptive procedure. Simulation results confirm the good prediction and the rankconsistency results under data-driven explicit choices of the tuning parameters and the scaling parameter of the noise.
Fichier principal
Vignette du fichier
2303.04694.pdf (431.39 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04419650 , version 1 (26-01-2024)

Identifiants

  • HAL Id : hal-04419650 , version 1

Citer

Nayel Bettache, Cristina Butucea. Two-sided Matrix Regression. 2024. ⟨hal-04419650⟩
31 Consultations
20 Téléchargements

Partager

Gmail Facebook X LinkedIn More