Reinforcement learning of coordination in cooperative multi-agent systems

S Kapetanakis; D Kudenko

Reinforcement learning of coordination in cooperative multi-agent systems

S Kapetanakis, D Kudenko

Computer Science

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

We report on an investigation of reinforcement learning techniques for the learning of coordination in cooperative multi-agent systems. Specifically, we focus on a novel action selection strategy for Q-learning (Watkins 1989). The new technique is applicable to scenarios where mutual observation of actions is not possible.

To date, reinforcement learning approaches for such independent agents did not guarantee convergence to the optimal joint action in scenarios with high miscoordination costs. We improve on previous results (Claus & Boutilier 1998) by, demonstrating empirically that our extension causes the agents to converge almost always to the optimal joint action even in these difficult cases.

Original language	English
Title of host publication	EIGHTEENTH NATIONAL CONFERENCE ON ARTIFICIAL INTELLIGENCE (AAAI-02)/FOURTEENTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE (IAAI-02), PROCEEDINGS
Place of Publication	CAMBRIDGE
Publisher	MIT Press
Pages	326-331
Number of pages	6
ISBN (Print)	0-262-51129-0
Publication status	Published - 2002
Event	AAAI-02 - Edmonton, Alberta, Canada Duration: 28 Jul 2002 → 1 Aug 2002

Conference

Conference	AAAI-02
Country/Territory	Canada
City	Edmonton, Alberta
Period	28/07/02 → 1/08/02

Cite this

@inproceedings{db937ebb4895489a9c82841a2fd7c771,

title = "Reinforcement learning of coordination in cooperative multi-agent systems",

abstract = "We report on an investigation of reinforcement learning techniques for the learning of coordination in cooperative multi-agent systems. Specifically, we focus on a novel action selection strategy for Q-learning (Watkins 1989). The new technique is applicable to scenarios where mutual observation of actions is not possible.To date, reinforcement learning approaches for such independent agents did not guarantee convergence to the optimal joint action in scenarios with high miscoordination costs. We improve on previous results (Claus & Boutilier 1998) by, demonstrating empirically that our extension causes the agents to converge almost always to the optimal joint action even in these difficult cases.",

author = "S Kapetanakis and D Kudenko",

year = "2002",

language = "English",

isbn = "0-262-51129-0",

pages = "326--331",

booktitle = "EIGHTEENTH NATIONAL CONFERENCE ON ARTIFICIAL INTELLIGENCE (AAAI-02)/FOURTEENTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE (IAAI-02), PROCEEDINGS",

publisher = "MIT Press ",

address = "United States",

note = "AAAI-02 ; Conference date: 28-07-2002 Through 01-08-2002",

}

Reinforcement learning of coordination in cooperative multi-agent systems. / Kapetanakis, S; Kudenko, D.
EIGHTEENTH NATIONAL CONFERENCE ON ARTIFICIAL INTELLIGENCE (AAAI-02)/FOURTEENTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE (IAAI-02), PROCEEDINGS. CAMBRIDGE: MIT Press , 2002. p. 326-331.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

TY - GEN

T1 - Reinforcement learning of coordination in cooperative multi-agent systems

AU - Kapetanakis, S

AU - Kudenko, D

PY - 2002

Y1 - 2002

N2 - We report on an investigation of reinforcement learning techniques for the learning of coordination in cooperative multi-agent systems. Specifically, we focus on a novel action selection strategy for Q-learning (Watkins 1989). The new technique is applicable to scenarios where mutual observation of actions is not possible.To date, reinforcement learning approaches for such independent agents did not guarantee convergence to the optimal joint action in scenarios with high miscoordination costs. We improve on previous results (Claus & Boutilier 1998) by, demonstrating empirically that our extension causes the agents to converge almost always to the optimal joint action even in these difficult cases.

AB - We report on an investigation of reinforcement learning techniques for the learning of coordination in cooperative multi-agent systems. Specifically, we focus on a novel action selection strategy for Q-learning (Watkins 1989). The new technique is applicable to scenarios where mutual observation of actions is not possible.To date, reinforcement learning approaches for such independent agents did not guarantee convergence to the optimal joint action in scenarios with high miscoordination costs. We improve on previous results (Claus & Boutilier 1998) by, demonstrating empirically that our extension causes the agents to converge almost always to the optimal joint action even in these difficult cases.

M3 - Conference contribution

SN - 0-262-51129-0

SP - 326

EP - 331

BT - EIGHTEENTH NATIONAL CONFERENCE ON ARTIFICIAL INTELLIGENCE (AAAI-02)/FOURTEENTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE (IAAI-02), PROCEEDINGS

PB - MIT Press

CY - CAMBRIDGE

T2 - AAAI-02

Y2 - 28 July 2002 through 1 August 2002

ER -