Voting policies that cope with unreliable agents

ABSTRACT

Collaboration plays a critical role when a team is striving for goals that are difficult to achieve by an individual. In previous work, we defined the ETAPP (Environment-Task-Agents-Policy-Protocol) framework, which describes the collaboration of a team of agents. According to this framework, team members propose agents to perform a task, and the team applies a voting policy to choose an agent for the task. In this paper, we expand on three parameters of this framework. We model team members that have variable proposal making attitudes, and team members whose performance exhibits different levels of stability. We then consider two new voting policies for group decision-making, and use a simulation-based evaluation to investigate the interaction between the different types of team members and the voting policies. Our results show that our previous optimistic voting policy, which chooses the agent that seems to have the best performance, yields an unstable task performance for teams where even a few agents do not make the best possible proposal. In contrast, our new voting policies yield a stable task performance.

REFERENCES

• 1: P. Busetta, M. Merzi, S. Rossi, and F. Legras. Intra-role coordination using group communication: A preliminary report. In F. Dignum, editor, Advances in Agent Communication, LNAI 2922. Springer, 2004.
• 2: V. Conitzer and T. Sandholm. Universal voting protocol tweaks to make manipulation hard. In IJCAI-2003 - Proceedings of the Eighteenth International Joint Conference on Artificial Intelligence, 781--788, Acapulco, Mexico, 2003.
• 3: B. Davison and H. Hirsh. Predicting sequences of user actions. In Notes of the AAAI/ICML 1998 Workshop on Predicting the Future: AI Approaches to Time-Series Analysis, Madison, Wisconsin, 1998.
• 4: Piotr J. Gmytrasiewicz , Edmund H. Durfee, Rational Communication in Multi-Agent Environments, Autonomous Agents and Multi-Agent Systems, v.4 n.3, p.233-272, September 2001 [doi>10.1023/A:1011495811107 ]
• 5: C. Guttmann and I. Zukerman. Towards Models of Incomplete and Uncertain Knowledge of Collaborators' Internal Resources. In J. Denzinger, G. Lindemann, I. J. Timm, and R. Unland, editors, Second German Conference on MultiAgent system TEchnologieS (MATES), LNAI 3187, Erfurt, Germany, 2004. Springer.
• 6: Lik Mui , Mojdeh Mohtashemi , Ari Halberstadt, Notions of reputation in multi-agents systems: a review, Proceedings of the first international joint conference on Autonomous agents and multiagent systems: part 1, July 15-19, 2002, Bologna, Italy [doi>10.1145/544741.544807]
• 7: David M. Pennock , Eric Horvitz , C. Lee Giles, Social Choice Theory and Recommender Systems: Analysis of the Axiomatic Foundations of Collaborative Filtering, Proceedings of the Seventeenth National Conference on Artificial Intelligence and Twelfth Conference on Innovative Applications of Artificial Intelligence, p.729-734, July 30-August 03, 2000
• 8: A. Rubinstein. Modeling bounded rationality. Zeuthen lecture book series. MIT Press, Cambridge, Massachusetts, 1998.
• 9: Dicky Suryadi , Piotr J. Gmytrasiewicz, Learning models of other agents using influence diagrams, Proceedings of the seventh international conference on User modeling, p.223-232, June 1999, Banff, Canada
• 10: M. Tambe. Towards Flexible Teamwork. Journal of Artificial Intelligence Research, 7:83--124, 1997.
• 11: Marilyn A. Walker, The effect of resource limits and task complexity on collaborative planning in dialogue, Artificial Intelligence, v.85 n.1-2, p.181-243, Aug. 1996 [doi>10.1016/0004-3702(95)00114-X ]

INDEX TERMS

Primary Classification:
  I. Computing Methodologies
  I.2 ARTIFICIAL INTELLIGENCE
        I.2.11 Distributed Artificial Intelligence
            Subjects: Multiagent systems

General Terms:
Design, Performance, Reliability

Keywords:
agent collaboration, unreliable team members, variable performance, voting policies