A Multi-Agent, Policy-Gradient approach to Network Routing

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Tao, Nigel, Baxter, Jonathan, Weaver, Lex
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866912745686827008
author Tao, Nigel
Baxter, Jonathan
Weaver, Lex
author_facet Tao, Nigel
Baxter, Jonathan
Weaver, Lex
contents Network routing is a distributed decision problem which naturally admits numerical performance measures, such as the average time for a packet to travel from source to destination. OLPOMDP, a policy-gradient reinforcement learning algorithm, was successfully applied to simulated network routing under a number of network models. Multiple distributed agents (routers) learned co-operative behavior without explicit inter-agent communication, and they avoided behavior which was individually desirable, but detrimental to the group's overall performance. Furthermore, shaping the reward signal by explicitly penalizing certain patterns of sub-optimal behavior was found to dramatically improve the convergence rate.
format Preprint
id arxiv_https___arxiv_org_abs_2512_03211
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle A Multi-Agent, Policy-Gradient approach to Network Routing
Tao, Nigel
Baxter, Jonathan
Weaver, Lex
Machine Learning
Networking and Internet Architecture
Network routing is a distributed decision problem which naturally admits numerical performance measures, such as the average time for a packet to travel from source to destination. OLPOMDP, a policy-gradient reinforcement learning algorithm, was successfully applied to simulated network routing under a number of network models. Multiple distributed agents (routers) learned co-operative behavior without explicit inter-agent communication, and they avoided behavior which was individually desirable, but detrimental to the group's overall performance. Furthermore, shaping the reward signal by explicitly penalizing certain patterns of sub-optimal behavior was found to dramatically improve the convergence rate.
title A Multi-Agent, Policy-Gradient approach to Network Routing
topic Machine Learning
Networking and Internet Architecture
url https://arxiv.org/abs/2512.03211