Multi-Agent Deep Reinforcement Learning for Collaborative Task Scheduling

Mali Gergely

2024

Abstract

Efficient, scalable and cost-efficient resource management is a multi-faceted online decision making problem, faced more and more in networking and cloud computing. More specifically, task-scheduling stands out as a complex challenge, solving which is critical for the optimal functioning of today’s systems. Traditional heuristic approaches to scheduling are laborious to design and especially difficult to tune, therefore various machine-learning based methods have been proposed. Reinforcement Learning (RL) showed great results in similar decision making problems, and many existing approaches employ RL to solve task scheduling problems. Most of these works consider either single-agent scenarios (and thus suffer from scalability issues), or the existing multi-agent applications are heavily specialised. We propose a general-purpose multi-agent RL framework that can successfully learn collaborative optimal scheduling policies, making one step further towards clouds and networks that are both scalable and autonomous. Our experiments show that these agents can collaboratively learn optimal scheduling policies for dynamic workloads.

Paper Citation

in Harvard Style

Gergely M. (2024). Multi-Agent Deep Reinforcement Learning for Collaborative Task Scheduling. In Proceedings of the 16th International Conference on Agents and Artificial Intelligence - Volume 3: ICAART; ISBN 978-989-758-680-4, SciTePress, pages 1076-1083. DOI: 10.5220/0012434700003636

in Bibtex Style

@conference{icaart24,
author={Mali Gergely},
title={Multi-Agent Deep Reinforcement Learning for Collaborative Task Scheduling},
booktitle={Proceedings of the 16th International Conference on Agents and Artificial Intelligence - Volume 3: ICAART},
year={2024},
pages={1076-1083},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0012434700003636},
isbn={978-989-758-680-4},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 16th International Conference on Agents and Artificial Intelligence - Volume 3: ICAART
TI - Multi-Agent Deep Reinforcement Learning for Collaborative Task Scheduling
SN - 978-989-758-680-4
AU - Gergely M.
PY - 2024
SP - 1076
EP - 1083
DO - 10.5220/0012434700003636
PB - SciTePress