期刊
JOURNAL OF STATISTICAL MECHANICS-THEORY AND EXPERIMENT
卷 2022, 期 8, 页码 -出版社
IOP Publishing Ltd
DOI: 10.1088/1742-5468/ac7e4c
关键词
learning theory; random matrix theory and extensions; statistical inference; spin glasses
资金
- ERC under the European Union's Horizon 2020 Research and Innovation Program [714608-SMiLe]
- 'Chaire de recherche sur les modeles et sciences des donnees', Fondation CFM pour la Recherche-ENS
The paper discusses matrix factorization and extensive-rank matrix denoising problems using high-temperature expansions to find more accurate solutions. It provides a systematic approach to derive corrections to existing approximations, taking into account the specific structure of correlations in the problems.
Factorization of matrices where the rank of the two factors diverges linearly with their sizes has many applications in diverse areas such as unsupervised representation learning, dictionary learning or sparse coding. We consider a setting where the two factors are generated from known component-wise independent prior distributions, and the statistician observes a (possibly noisy) component-wise function of their matrix product. In the limit where the dimensions of the matrices tend to infinity, but their ratios remain fixed, we expect to be able to derive closed form expressions for the optimal mean squared error on the estimation of the two factors. However, this remains a very involved mathematical and algorithmic problem. A related, but simpler, problem is extensive-rank matrix denoising, where one aims to reconstruct a matrix with extensive but usually small rank from noisy measurements. In this paper, we approach both these problems using high-temperature expansions at fixed order parameters. This allows to clarify how previous attempts at solving these problems failed at finding an asymptotically exact solution. We provide a systematic way to derive the corrections to these existing approximations, taking into account the structure of correlations particular to the problem. Finally, we illustrate our approach in detail on the case of extensive-rank matrix denoising. We compare our results with known optimal rotationally-invariant estimators, and show how exact asymptotic calculations of the minimal error can be performed using extensive-rank matrix integrals.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据