Strategic Coalition for Improving Generalization Ability of Multi-agent with Evolutionary Learning

진화학습을 이용한 다중에이전트의 일반화 성능향상을 위한 전략적 연합

  • Published : 2004.02.01

Abstract

In dynamic systems, such as social and economic systems, complex interactions emerge among its members. In that case, their behaviors become adaptive according to Changing environment. In many cases, an individual's behaviors can be modeled by a stimulus-response system in a dynamic environment. In this paper, we use the Iterated Prisoner's Dilemma (IPD) game, which is simple yet capable of dealing with complex problems, to model the dynamic systems. We propose strategic coalition consisting of many agents and simulate their emergence in a co-evolutionary learning environment. Also we introduce the concept of confidence for agents in a coalition and show how such confidences help to improve the generalization ability of the whole coalition. Experimental results are presented to demonstrate that co-evolutionary learning with coalitions and confidence allows better performing strategies that generalize well.

사회시스템이나 경제시스템 같이 동적으로 변하는 시스템에서는 그 구성원들 간에 복잡한 상호작용(행동)이 나타나게 되는데 구성원들의 행동은 변화하는 환경에 따라 적응하는 경향을 보인다. 그리고 이들의 행동양상은 흔히 생물학 분야의 조건반사에 비유되기도 한다. 본 논문에서는 복잡한 사회 현상을 모델링하고 분석하기 위하여 반복적 죄수의 딜레마 게임 상에서 에이전트들의 전략적 연합을 통하여 변화하는 환경에 잘 적응하는 일반화 능력이 우수한 에이전트들을 자동 생성하는 방법을 제안한다. 또한 에이전트에 신뢰도를 부여하여 연하의 의사결정에 참가하게 함으로써 일반화 성능을 향상시키는 방법을 소개한다 실험결과, 전략적 연합을 이용하여 진화된 에이전트들은 테스트 에이전트들에 비하여 일반화 성능이 우수함을 확인할 수 있었다.

Keywords

References

  1. M. Patrignani, 'Stability of arbitrary genes: A new approach to cooperation,' Proceedings of Genetic and Evolutionary Computation Conference, Morgan Kaufmann, San Francisco, California, USA, pp. 907, 2001
  2. A. Francisco, 'A computational evolutionary approach to evolving game strategy and cooperation,' IEEE Transactions on Systems, Man and Cybernetics, Part B, Vol. 32, No. 5, pp. 498-502, 2002 https://doi.org/10.1109/TSMCB.2003.810948
  3. 서연규, 조성배, '진화방식을 이용한 N명 죄수 딜레마 게임의 협동연합에 관한 실험적 연구', 한국정보과학회 논문지 (B), Vol. 27(3), pp. 257-265, 2000
  4. A. M. Colman, Game Theory and Experimental Games, Oxford, England: Pergamon Press, 1982
  5. T. Ord and A. Blair, 'Exploitation and peacekeeping: Introducing more sophisticated interactions to the iterated prisoner's dilemma.' Proceedings of 2002 Congress on Evolutionary Computation, Vol. 2, pp. 1606-1611, 2002 https://doi.org/10.1109/CEC.2002.1004482
  6. L.Tesfatsion, 'Agent-based computational economics: Growing economics from the bottom up,' Artificial Life, Vol. 8, pp. 55-82, 2002 https://doi.org/10.1162/106454602753694765
  7. X. Yao and P. J. Darwen, 'An experimental study of N-person iterated prisoner's dilemma games,' Informatica, Vol. 18, pp. 435-450, 1994
  8. Y. G. Seo, S. B. Cho and X. Yao, 'Exploiting coalition in co-evolutionary learning,' Proceedings of Congress on Evolutionary Computation 2000, Vol. 2, pp. 1268-1275, 2000 https://doi.org/10.1109/CEC.2000.870796
  9. J. A. Fletcher and M. Zwick, 'N-Player prisoner's dilemma in multiple groups: A model of multilevel selection,' Proceedings of Artificial Life VII Workshops, Portland, Oregon, 2000
  10. R. Axelrod, The Evolution of Cooperations, New York: Basic Books, 1984
  11. J. Nash, 'Equilibrium points in n-person games,' Proceedings of National Academy of Sciences, Vol. 36, pp. 48-49, 1950 https://doi.org/10.1073/pnas.36.1.48
  12. D. Ashlock and M. Joenks, 'ISAc lists, a different representation for program induction,' Genetic Programming 98, Proceeding of the Third Annual Genetic Programming Conference, pp. 3-10, San Francisco, Morgan Kaufmann, 1998
  13. P. J. Darwen and X. Yao, 'Speciation as automatic categorical modularization,' IEEE Transactions on Evolutionary Computation, Vol. 1, No. 2, pp. 101-108, 1997 https://doi.org/10.1109/4235.687878
  14. K. P. Sycara, 'Persuasive argumentation in negotiation,' Theory and Decision, Vol. 28, pp. 203-242, 1990 https://doi.org/10.1007/BF00162699
  15. G. Zlotkin and J. Rosenschein, 'Cooperation and conflict resolution via negotiation among autonomous agents in noncooperative domains,' IEEE Transactions on Systems, Man, and Cybernetics, 21(6), pp. 1317-1324, 1991 https://doi.org/10.1109/21.135678
  16. O. Shehory and S. Kraus, 'Coalition formation among autonomous agents: Strategies and complexity,' Fifth European Workshop on Modeling Autonomous Agents in a Multi-Agent World, Springer-Verlag, Heidelberg, Germany, pp. 56-72, 1993
  17. T. W. Sandholm and V. R. Lesser, 'Coalitions among computationally bounded agents,' Artificial Intelligence, Vol. 94, pp. 99-137, 1997 https://doi.org/10.1016/S0004-3702(97)00030-1