Saved in:
Bibliographic Details
Main Authors: Grant, Curtis, Jagannath, Aukosh, Ko, Justin
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2503.01708
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915603376242688
author Grant, Curtis
Jagannath, Aukosh
Ko, Justin
author_facet Grant, Curtis
Jagannath, Aukosh
Ko, Justin
contents We develop a pseudo-likelihood theory for rank one matrix estimation problems in the high dimensional limit. We prove a variational principle for the limiting pseudo-maximum likelihood which also characterizes the performance of the corresponding pseudo-maximum likelihood estimator. We show that this variational principle is universal and depends only on four parameters determined by the corresponding null model. Through this universality, we introduce a notion of equivalence for estimation problems of this type and, in particular, show that a broad class of estimation tasks, including community detection, sparse submatrix detection, and non-linear spiked matrix models, are equivalent to spiked matrix models. As an application, we obtain a complete description of the performance of the least-squares (or ``best rank one'') estimator for any rank one matrix estimation problem.
format Preprint
id arxiv_https___arxiv_org_abs_2503_01708
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Pseudo-Maximum Likelihood Theory for High-Dimensional Rank One Inference
Grant, Curtis
Jagannath, Aukosh
Ko, Justin
Statistics Theory
Probability
We develop a pseudo-likelihood theory for rank one matrix estimation problems in the high dimensional limit. We prove a variational principle for the limiting pseudo-maximum likelihood which also characterizes the performance of the corresponding pseudo-maximum likelihood estimator. We show that this variational principle is universal and depends only on four parameters determined by the corresponding null model. Through this universality, we introduce a notion of equivalence for estimation problems of this type and, in particular, show that a broad class of estimation tasks, including community detection, sparse submatrix detection, and non-linear spiked matrix models, are equivalent to spiked matrix models. As an application, we obtain a complete description of the performance of the least-squares (or ``best rank one'') estimator for any rank one matrix estimation problem.
title Pseudo-Maximum Likelihood Theory for High-Dimensional Rank One Inference
topic Statistics Theory
Probability
url https://arxiv.org/abs/2503.01708