Asymptotic Dynamics of Alternating Minimization for Bilinear Regression

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Okajima, Koki, Takahashi, Takashi
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929692797304832
author Okajima, Koki
Takahashi, Takashi
author_facet Okajima, Koki
Takahashi, Takashi
contents This study investigates the dynamics of alternating minimization applied to a bilinear regression task with normally distributed covariates, under the asymptotic system size limit where the number of parameters and observations diverge at the same rate. This is achieved by employing the replica method to a multi-temperature glassy system which unfolds the algorithm's time evolution. Our results show that the dynamics can be described effectively by a two-dimensional discrete stochastic process, where each step depends on all previous time steps, revealing the structure of the memory dependence in the evolution of alternating minimization. The theoretical framework developed in this work can be applied to the analysis of various iterative algorithms, extending beyond the scope of alternating minimization.
format Preprint
id arxiv_https___arxiv_org_abs_2402_04751
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Asymptotic Dynamics of Alternating Minimization for Bilinear Regression
Okajima, Koki
Takahashi, Takashi
Optimization and Control
Disordered Systems and Neural Networks
Machine Learning
This study investigates the dynamics of alternating minimization applied to a bilinear regression task with normally distributed covariates, under the asymptotic system size limit where the number of parameters and observations diverge at the same rate. This is achieved by employing the replica method to a multi-temperature glassy system which unfolds the algorithm's time evolution. Our results show that the dynamics can be described effectively by a two-dimensional discrete stochastic process, where each step depends on all previous time steps, revealing the structure of the memory dependence in the evolution of alternating minimization. The theoretical framework developed in this work can be applied to the analysis of various iterative algorithms, extending beyond the scope of alternating minimization.
title Asymptotic Dynamics of Alternating Minimization for Bilinear Regression
topic Optimization and Control
Disordered Systems and Neural Networks
Machine Learning
url https://arxiv.org/abs/2402.04751