Saved in:
Bibliographic Details
Main Authors: Shustova, Evgenia, Sheshukova, Marina, Samsonov, Sergey, Frolov, Evgeny
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2510.19349
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908821696282624
author Shustova, Evgenia
Sheshukova, Marina
Samsonov, Sergey
Frolov, Evgeny
author_facet Shustova, Evgenia
Sheshukova, Marina
Samsonov, Sergey
Frolov, Evgeny
contents In this paper, we introduce PSI-LinUCB, a scalable variant of LinUCB that enables efficient training, inference, and memory usage by representing the inverse regularized design matrix as a sum of a diagonal matrix and low-rank correction. We derive numerically stable rank-1 and batched updates that maintain the inverse without explicitly forming the matrix. To control memory growth, we employ a projector-splitting integrator for dynamical low-rank approximation, yielding an average per-step update cost and memory usage of $O(dr)$ for approximation rank $r$. The inference complexity of the proposed algorithm is $O(dr)$ per action evaluation. Experiments on recommender system datasets demonstrate the effectiveness of our algorithm.
format Preprint
id arxiv_https___arxiv_org_abs_2510_19349
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Scalable LinUCB: Low-Rank Design Matrix Updates for Recommenders with Large Action Spaces
Shustova, Evgenia
Sheshukova, Marina
Samsonov, Sergey
Frolov, Evgeny
Machine Learning
68T05, 65F30
In this paper, we introduce PSI-LinUCB, a scalable variant of LinUCB that enables efficient training, inference, and memory usage by representing the inverse regularized design matrix as a sum of a diagonal matrix and low-rank correction. We derive numerically stable rank-1 and batched updates that maintain the inverse without explicitly forming the matrix. To control memory growth, we employ a projector-splitting integrator for dynamical low-rank approximation, yielding an average per-step update cost and memory usage of $O(dr)$ for approximation rank $r$. The inference complexity of the proposed algorithm is $O(dr)$ per action evaluation. Experiments on recommender system datasets demonstrate the effectiveness of our algorithm.
title Scalable LinUCB: Low-Rank Design Matrix Updates for Recommenders with Large Action Spaces
topic Machine Learning
68T05, 65F30
url https://arxiv.org/abs/2510.19349