We extend the study of congestion problems to a more realistic scenario, the Road Network Domain (RND), where the resources are no longer independent, but rather part of a network, thus choosing one path will also impact the load of another one having common road segments. We demonstrate the application of state-of-the-art multi-agent reinforcement learning methods for this new congestion model and analyse their performance. RND allows us to highlight an important limitation of resource abstraction and show that the difference rewards approach manages to better capture and inform the agents about the dynamics of the environment.
Radulescu, R, Vrancx, P & Nowe, A 2017, Analysing Congestion Problems in Multi-agent Reinforcement Learning. in E Durfee, M Winikoff, K Larson & S Das (eds), 16th International Conference on Autonomous Agents and Multiagent Systems, AAMAS 2017. vol. 3, pp. 1705-1707, 16th International Conference on Autonomous Agents and Multiagent Systems, Sao Paolo, Brazil, 8/05/17. <http://www.ifaamas.org/Proceedings/aamas2017/pdfs/p1705.pdf>
Radulescu, R., Vrancx, P., & Nowe, A. (2017). Analysing Congestion Problems in Multi-agent Reinforcement Learning. In E. Durfee, M. Winikoff, K. Larson, & S. Das (Eds.), 16th International Conference on Autonomous Agents and Multiagent Systems, AAMAS 2017 (Vol. 3, pp. 1705-1707) http://www.ifaamas.org/Proceedings/aamas2017/pdfs/p1705.pdf
@inproceedings{91473affc5b34b308aec088a1e4a452b,
title = "Analysing Congestion Problems in Multi-agent Reinforcement Learning",
abstract = "We extend the study of congestion problems to a more realistic scenario, the Road Network Domain (RND), where the resources are no longer independent, but rather part of a network, thus choosing one path will also impact the load of another one having common road segments. We demonstrate the application of state-of-the-art multi-agent reinforcement learning methods for this new congestion model and analyse their performance. RND allows us to highlight an important limitation of resource abstraction and show that the difference rewards approach manages to better capture and inform the agents about the dynamics of the environment.",
keywords = "Congestion problems, Multi-agent reinforcement learning, Resource abstraction",
author = "Roxana Radulescu and Peter Vrancx and Ann Nowe",
year = "2017",
month = may,
day = "8",
language = "English",
volume = "3",
pages = "1705--1707",
editor = "Edmund Durfee and Michael Winikoff and Kate Larson and Sanmay Das",
booktitle = "16th International Conference on Autonomous Agents and Multiagent Systems, AAMAS 2017",
note = "16th International Conference on Autonomous Agents and Multiagent Systems : AAMAS 2017, AAMAS 2017 ; Conference date: 08-05-2017 Through 12-05-2017",
url = "http://www.aamas2017.org",
}