Learning to Coordinate Using Commitment Sequences in Cooperative Multi-agent Systems

Spiros Kapetanakis; Daniel Kudenko; Malcolm J. A. Strens

Learning to Coordinate Using Commitment Sequences in Cooperative Multi-agent Systems

Spiros Kapetanakis, Daniel Kudenko, Malcolm J. A. Strens

Computer Science

Research output: Contribution to conference › Paper › peer-review

Abstract

We report on an investigation of the learning of coordination in cooperative multiagent systems. Specifically, we study solutions that are applicable to independent agents, i.e., agents that do not observe one another’s actions and do not explicitly communicate with each other. In previously published work (Kapetanakis and Kudenko, 2002) we have presented a reinforcement learning approach that converges to the optimal joint action even in scenarios with high miscoordination costs. However, this approach failed in fully stochastic environments. In this paper, we present a novel approach based on reward estimation with a shared action-selection protocol. The new technique is applicable in fully stochastic environments where mutual observation of actions is not possible. We demonstrate empirically that our approach causes the agents to converge almost always to the optimal joint action even in difficult stochastic scenarios with high miscoordination penalties

Original language	Undefined/Unknown
Pages	106-118
Publication status	Published - 2005

Access to Document

Cite this

@conference{393702f4dd584729a534489d617c5083,

title = "Learning to Coordinate Using Commitment Sequences in Cooperative Multi-agent Systems",

abstract = "We report on an investigation of the learning of coordination in cooperative multiagent systems. Specifically, we study solutions that are applicable to independent agents, i.e., agents that do not observe one another{\textquoteright}s actions and do not explicitly communicate with each other. In previously published work (Kapetanakis and Kudenko, 2002) we have presented a reinforcement learning approach that converges to the optimal joint action even in scenarios with high miscoordination costs. However, this approach failed in fully stochastic environments. In this paper, we present a novel approach based on reward estimation with a shared action-selection protocol. The new technique is applicable in fully stochastic environments where mutual observation of actions is not possible. We demonstrate empirically that our approach causes the agents to converge almost always to the optimal joint action even in difficult stochastic scenarios with high miscoordination penalties",

author = "Spiros Kapetanakis and Daniel Kudenko and Strens, {Malcolm J. A.}",

year = "2005",

language = "Undefined/Unknown",

pages = "106--118",

}

TY - CONF

T1 - Learning to Coordinate Using Commitment Sequences in Cooperative Multi-agent Systems

AU - Kapetanakis, Spiros

AU - Kudenko, Daniel

AU - Strens, Malcolm J. A.

PY - 2005

Y1 - 2005

N2 - We report on an investigation of the learning of coordination in cooperative multiagent systems. Specifically, we study solutions that are applicable to independent agents, i.e., agents that do not observe one another’s actions and do not explicitly communicate with each other. In previously published work (Kapetanakis and Kudenko, 2002) we have presented a reinforcement learning approach that converges to the optimal joint action even in scenarios with high miscoordination costs. However, this approach failed in fully stochastic environments. In this paper, we present a novel approach based on reward estimation with a shared action-selection protocol. The new technique is applicable in fully stochastic environments where mutual observation of actions is not possible. We demonstrate empirically that our approach causes the agents to converge almost always to the optimal joint action even in difficult stochastic scenarios with high miscoordination penalties

AB - We report on an investigation of the learning of coordination in cooperative multiagent systems. Specifically, we study solutions that are applicable to independent agents, i.e., agents that do not observe one another’s actions and do not explicitly communicate with each other. In previously published work (Kapetanakis and Kudenko, 2002) we have presented a reinforcement learning approach that converges to the optimal joint action even in scenarios with high miscoordination costs. However, this approach failed in fully stochastic environments. In this paper, we present a novel approach based on reward estimation with a shared action-selection protocol. The new technique is applicable in fully stochastic environments where mutual observation of actions is not possible. We demonstrate empirically that our approach causes the agents to converge almost always to the optimal joint action even in difficult stochastic scenarios with high miscoordination penalties

M3 - Paper

SP - 106

EP - 118

ER -