Saved in:
Bibliographic Details
Main Authors: Capponi, Agostino, Stojnic, Mihailo
Format: Preprint
Published: 2023
Subjects:
Online Access:https://arxiv.org/abs/2401.00578
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914626096070656
author Capponi, Agostino
Stojnic, Mihailo
author_facet Capponi, Agostino
Stojnic, Mihailo
contents We study the completion of approximately low rank matrices with entries missing not at random (MNAR). In the context of typical large-dimensional statistical settings, we establish a framework for the performance analysis of the nuclear norm minimization ($\ell_1^*$) algorithm. Our framework produces \emph{exact} estimates of the worst-case residual root mean squared error and the associated phase transitions (PT), with both exhibiting remarkably simple characterizations. Our results enable to {\it precisely} quantify the impact of key system parameters, including data heterogeneity, size of the missing block, and deviation from ideal low rankness, on the accuracy of $\ell_1^*$-based matrix completion. To validate our theoretical worst-case RMSE estimates, we conduct numerical simulations, demonstrating close agreement with their numerical counterparts.
format Preprint
id arxiv_https___arxiv_org_abs_2401_00578
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Exact Error in Matrix Completion: Approximately Low-Rank Structures and Missing Blocks
Capponi, Agostino
Stojnic, Mihailo
Information Theory
62B10, 94A16, 62D10
We study the completion of approximately low rank matrices with entries missing not at random (MNAR). In the context of typical large-dimensional statistical settings, we establish a framework for the performance analysis of the nuclear norm minimization ($\ell_1^*$) algorithm. Our framework produces \emph{exact} estimates of the worst-case residual root mean squared error and the associated phase transitions (PT), with both exhibiting remarkably simple characterizations. Our results enable to {\it precisely} quantify the impact of key system parameters, including data heterogeneity, size of the missing block, and deviation from ideal low rankness, on the accuracy of $\ell_1^*$-based matrix completion. To validate our theoretical worst-case RMSE estimates, we conduct numerical simulations, demonstrating close agreement with their numerical counterparts.
title Exact Error in Matrix Completion: Approximately Low-Rank Structures and Missing Blocks
topic Information Theory
62B10, 94A16, 62D10
url https://arxiv.org/abs/2401.00578