Asynchronous Policy Gradient Aggregation for Efficient Distributed Reinforcement Learning

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Tyurin, Alexander, Spiridonov, Andrei, Rudenko, Varvara
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866918413781172224
author Tyurin, Alexander
Spiridonov, Andrei
Rudenko, Varvara
author_facet Tyurin, Alexander
Spiridonov, Andrei
Rudenko, Varvara
contents We study distributed reinforcement learning (RL) with policy gradient methods under asynchronous and parallel computations and communications. While non-distributed methods are well understood theoretically and have achieved remarkable empirical success, their distributed counterparts remain less explored, particularly in the presence of heterogeneous asynchronous computations and communication bottlenecks. We introduce two new algorithms, Rennala NIGT and Malenia NIGT, which implement asynchronous policy gradient aggregation and achieve state-of-the-art efficiency. In the homogeneous setting, Rennala NIGT provably improves the total computational and communication complexity while supporting the AllReduce operation. In the heterogeneous setting, Malenia NIGT simultaneously handles asynchronous computations and heterogeneous environments with strictly better theoretical guarantees. Our results are further corroborated by experiments, showing that our methods significantly outperform prior approaches.
format Preprint
id arxiv_https___arxiv_org_abs_2509_24305
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Asynchronous Policy Gradient Aggregation for Efficient Distributed Reinforcement Learning
Tyurin, Alexander
Spiridonov, Andrei
Rudenko, Varvara
Machine Learning
Distributed, Parallel, and Cluster Computing
Optimization and Control
We study distributed reinforcement learning (RL) with policy gradient methods under asynchronous and parallel computations and communications. While non-distributed methods are well understood theoretically and have achieved remarkable empirical success, their distributed counterparts remain less explored, particularly in the presence of heterogeneous asynchronous computations and communication bottlenecks. We introduce two new algorithms, Rennala NIGT and Malenia NIGT, which implement asynchronous policy gradient aggregation and achieve state-of-the-art efficiency. In the homogeneous setting, Rennala NIGT provably improves the total computational and communication complexity while supporting the AllReduce operation. In the heterogeneous setting, Malenia NIGT simultaneously handles asynchronous computations and heterogeneous environments with strictly better theoretical guarantees. Our results are further corroborated by experiments, showing that our methods significantly outperform prior approaches.
title Asynchronous Policy Gradient Aggregation for Efficient Distributed Reinforcement Learning
topic Machine Learning
Distributed, Parallel, and Cluster Computing
Optimization and Control
url https://arxiv.org/abs/2509.24305