Enregistré dans:
Détails bibliographiques
Auteurs principaux: Glowacki, Bartosz, Kulik, Rafal, Soulier, Philippe
Format: Preprint
Publié: 2026
Sujets:
Accès en ligne:https://arxiv.org/abs/2605.07184
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866915991575855104
author Glowacki, Bartosz
Kulik, Rafal
Soulier, Philippe
author_facet Glowacki, Bartosz
Kulik, Rafal
Soulier, Philippe
contents Stochastic gradient descent (SGD) with mini-batching is a standard tool in large-scale optimization, yet its theoretical properties under heavy-tailed gradient noise remain largely unexplored. In this paper we study SGD with increasing batch sizes when the gradient noise belongs to the domain of attraction of an $α$-stable law with $α\in(1,2)$. Building on existing results for the finite-variance regime and for heavy-tailed SGD without batching, we establish three main results. First, we derive $L^p$ moment bounds for the SGD error and show that increasing batch sizes lead to faster convergence rates. In particular, batching enables convergence in probability even for a constant stepsize. Second, we prove that the properly normalized SGD iterates converge in distribution to the stationary law of an Ornstein-Uhlenbeck process driven by an $α$-stable Lévy process. Third, for Polyak-Ruppert averaging we obtain a stable limit theorem with a normalization that explicitly depends on the batch-size schedule.
format Preprint
id arxiv_https___arxiv_org_abs_2605_07184
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Convergence of Stochastic Gradient Descent with mini-batching and infinite variance
Glowacki, Bartosz
Kulik, Rafal
Soulier, Philippe
Probability
Stochastic gradient descent (SGD) with mini-batching is a standard tool in large-scale optimization, yet its theoretical properties under heavy-tailed gradient noise remain largely unexplored. In this paper we study SGD with increasing batch sizes when the gradient noise belongs to the domain of attraction of an $α$-stable law with $α\in(1,2)$. Building on existing results for the finite-variance regime and for heavy-tailed SGD without batching, we establish three main results. First, we derive $L^p$ moment bounds for the SGD error and show that increasing batch sizes lead to faster convergence rates. In particular, batching enables convergence in probability even for a constant stepsize. Second, we prove that the properly normalized SGD iterates converge in distribution to the stationary law of an Ornstein-Uhlenbeck process driven by an $α$-stable Lévy process. Third, for Polyak-Ruppert averaging we obtain a stable limit theorem with a normalization that explicitly depends on the batch-size schedule.
title Convergence of Stochastic Gradient Descent with mini-batching and infinite variance
topic Probability
url https://arxiv.org/abs/2605.07184