Download PDF (external access)

11th International Workshop on Cooperative Information Agents Delft, NETHERLANDS, SEP 19-21, 2007, Date: 2007/09/19 - 2007/09/21, Location: NETHERLANDS, Delft

Publication date: 2007-01-01
Volume: 4676 Pages: 36 - 56
ISSN: 978-3-540-75118-2
Publisher: Springer-verlag berlin; HEIDELBERGER PLATZ 3, D-14197 BERLIN, GERMANY

Cooperative Information Agents XI, Proceedings

Author:

van den Herik, H Jaap
Hennes, D ; Kaisers, M ; Tuyls, K ; Verbeeck, Katja

Keywords:

automata, Science & Technology, Technology, Computer Science, Artificial Intelligence, Computer Science, Hardware & Architecture, Computer Science, Information Systems, Computer Science, AUTOMATA

Abstract:

In this paper we compare state-of-the-art multi-agent reinforcement learning algorithms in a wide variety of games. We consider two types of algorithms: value iteration and policy iteration. Four characteristics are studied: initial conditions, parameter settings, convergence speed, and local versus global convergence. Global convergence is still difficult to achieve in practice, despite existing theoretical guarantees. Multiple visualizations are included to provide a comprehensive insight into the learning dynamics.