6期 金士尧等:面向涌现的多Agent系统研究及其进展 889 Pouhon等人l8 钉]通过这种方法,实现了二维 表3 De Wolf等人的模式描述框架 积木的自组装.他们将“涌现媒介”称为“酶”.“酶”是 描述要素 含义 由积木按照一定规则相互作用、通过涌现产生的、由 模式名/别名 指称解决方案或有用隐喻的确 若干积木构成的稳定结构.分析目标结构可以确定 (Pattern Name/Also Known As) 定清晰的名称和别名 需要什么样的“酶”,通过特定的“酶”可以获得目标 上下文/适用性 指出解决方案的应用场景,特 (Context/Applicability) 别何时该自组织涌现解决方案 积木结构.而相应的“酶”可以通过控制作用规则,由 最有前景 积木自底向上涌现产生. 针对问题/意图 指出该模式解决什么问题.工程 (Problem/Intent) 师通过比较其问题来选择模式 (4)仿真在分析设计中的重要作用 约束(Forces) 为一个问题选择解决方案时, 测试与验证是保证所设计的系统满足目标的关 需要考虑的对立因素 解决方案(Solution) 描述怎样解决给出的问题 键一环.在面向涌现的分析设计中,测试与验证工作 相关模式/机制 相关的一些模式以及它们的异 非常重要的一个方面,是确定是否能够获得期望的 (Related Mechanisms/Patterns) 同点 涌现特征.由于涌现特征是在系统范围存在并动态 实例/已知应用 案例形式给出的已知应用的 (Examples/Known Uses) 实例 保持的,传统的测试验证方法(单元测试、场景测试 等)不能确定其是否满足目标.另外,MAS是一个复 Gardelli等人认为De Wolf等人给出的模式是 杂的交互计算系统,要建立起一个有效的形式化模 一种复杂的模式,面它们可以通过更简单更基本的 型来验证其正确性非常困难.正是由于这些原因,使 模式构建出来 .他们给出的MAS自组织设计模 得仿真成为一种不可或缺的测试与验证方法[7 . 式及其描述框架如表4和表5.其中,描述框架采用 De Wolf等人_8 使用基于仿真的科学分析方 的是Lind提出的框架 . 法,来确定MAS是否具有期望的宏观涌现行为. 表4 Gardelli等人给出的自组织设计模式 Gardelli等 胡则直接使用仿真来发现设计缺陷,验 模式 针对问题 证系统的正确性. 复制 (1)降低信息的访问时间 5.2设计模式 (Replication) (2)在受攻击或失效时,避免信息丢失 设计模式_8 作为一种重要的软件重用方式而 联合分类 在没有对信息仓库做任何显式限定时,MAS (Collective Sort) 环境查找信息的开销可能增大 受到设计者们的青睐.面向涌现的MAS设计模式, 耀件 , . 、 MAS环境会被Agent放置的过多信息所淹没 试图将MAS涌现性设计经验以规范化的描述记录 Lvaporazion) 聚合 大规模的MAS为综合出宏观信息,必须筛 下来,以获得成熟设计经验的重用.这首先需要从已 (Aggregation) 选Agent放置的大量信息,这耗费较多资源 有的应用中甄别出一些固定的模式,然后以确定的 扩散 MAS中Agent只能访问局部信息,Agent推 描述框架描述出来.这里我们主要介绍De Wolf等 (Diffusion) 理缺乏关于邻居的知识 人_8[书 以及Gardelli等人 在MAS自组织设计模 表5 Gardelli等人的模式描述框架 式方面的工作. 描述要素 含义 De Wolf等人_8 书 从MAS自组织应用中甄别 模式名(Name) 模式的名称 出的自组织协调模式如表2,所使用的描述框架如 别名(Aliases) 模式名的别名 针对问题(Problem) 该模式所解决的问题 表3.这些模式主要受显示自然、社会和人工系统的 约束(Forces) 衡量标准之间的权衡 启发而得到. 参与实体(Entities) 参与到该模式中的实体 动力学(Dynamics) 实体之间的交互 表2 De Wolf等人给出的自组织协调模式 依赖(Dependencies) 环境需求 实例(Example) 模式 针对问题 模式使用的一个抽象例子 实现(Implementation) 对模式实现的提示 (Dig h P字信息素路 已知应用(Known Uses) 使用该模式的现有应用 ;+ heromoneP、 ; ’ 譬譬影响(consequences) 对整个系统设计的作用 参考(See Also) 给出共参考的模式 梯度场 空间运动;模式生成;结构生成; (Gradient Fields) 路由;上下文信息的整合 基于市场的协调 从De WoK等人和Gardelli等人的研究可以看 一般的资源分配;资源使用或需 (Market-based Coordination) 要信息的整合 出,面向涌现的MAS设计模式是MAS涌现机制及 标签 (Tags) 团队形成;信任和声誉 其用于解决有关问题的规范描述,只不过他们是分 令牌 别从不同的层次给出的.De WoK等人及Gardelli (Tokens) 同步;资源分配 等人给出的只是MAS自组织的设计模式,其它涌 维普资讯 http://www.cqvip.com 890 计 算 机 学 报 现特征的设计模式还有待于进一步的发掘. 信息经济中,控制避免大规模的价格波动涌现 发生E叽]. 6问题与方向 面向涌现的MAS研究正方兴未艾,取得了一 控制论告诉我们,一旦知道所有参数和系统演 化规律,控制活动就归结为改变操作参数(opera— tional parameters)。所以,在涌现性度量的基础上, 些阶段性成果,同时存在一些亟待解决的问题,在一 些方向上需要更进一步的研究. MAS涌现性的控制,一定程度上,就是确保相关度 量值在相应的范围内.但是这并不是件简单事.因 为,在大多情况下,我们并不知道MAS的操作参数 和支配规律;并且,即使知道这些参数和规律,有时 仍然缺乏影响系统行为的Agent微观控制手段. 因此,要对MAS的涌现特征进行有效控制,还 需对支配MAS演化的规律进行研究以及探究系统 的微~宏观控制手段. (1)基于逻辑的涌现性宏观描述的完善 从MAS研究开始,人们就一直偏爱用逻辑来 描述MAS.基于逻辑对MAS的一些涌现特征进行 描述,是一种崭新的尝试.Dunin-Keplicz等人给出 的群体思维状态逻辑模型_2。 ,在MAS群体思维 状态的涌现性描述方面进行了有用的探索,具有非 常重要的意义.因为MAS联合思维状态并非多个 Agent的思维状态的简单联合,在一定程度上具有 涌现性.但是,这仅仅是一个开始.用逻辑对其它一 (4)MAS涌现的微一宏观机制的动力学研究 针对MAS涌现性控制存在的问题,一种重要 解决方案是在MAS中引入成熟的涌现机制.如果 我们对这些微一宏观机制产生的涌现行为有比较全 面的认识,就能对MAS的涌现特征进行有效控制. 对MAS涌现的微一宏观机制进行动力学研究, 能够帮助我们了解MAS的演化规律和发现微一宏 观控制手段.虽然受生物、经济和社会等系统的启 发,人们甄别出了许多有用的涌现机制,但是在对它 些涌现特征进行描述,还有待进一步研究.并且 Dunin—Keplicz等人研究的MAS群体思维状态是基 于BDI 9_逻辑的,利用其它逻辑模型,例如Brafman 和Tennenholtz的基于信念、偏好和决策策略Agent 逻辑模型llg ,会得到什么结果呢?这也是可以进一 步研究的问题. (2)MAS涌现性的度量 系统度量是认识一个系统的必要活动.MAS涌 们进行定量的动力学研究上,还只是个案研究(如博 弈机制_3 。 ),有待进一步全面展开. (5)MAS面向涌现的设计方法和设计模式的 完善 对于一些MAS而言,不可避免地存在一些涌 现性的度量是认识、设计和控制MAS的涌现性的 关键一环.涌现性是针对系统的宏观特征而言的.要 在宏观层面上度量一个系统,需要引入一些综合指 标,以反映系统的相关特征.在此基础上,也能对两 个系统的行为,或同一系统在不同时间、不同条件下 的行为,进行比较. 用“熵”l_】 。。。幻来度量MAS的一些涌现特征, 现特征.怎样化弊为利,发掘涌现的作用?除了涌现 性控制外,非常重要的一点是在分析时充分考虑 MAS的涌现性要求,并在设计时采取相应机制保证 (或避免)一些涌现特征的出现,或设计相应控制手 确实可以得到一些重要结论.但是在原始意义上, “熵”仅是系统无序度的度量,不能反映所有的涌现 特征.所以,我们必须确定MAS中的哪些宏观参数 需要测量,以便度量系统的涌现性以及反映系统的 一段以便运行时控制涌现现象的发生.这就要借助面 向涌现的MAS设计. MAS面向涌现的设计方法,目前大多是在传统 方法中加入涌现性设计步骤,并且主要围绕MAS 自组织展开L7 .面向涌现的设计模式也是针对 MAS自组织的 。 .这些都难以满足MAS的涌现 般本质.在这方面,一个比较有效的途径,是从所 有涉及宏观度量的学科中获得启发和隐喻,例如,热 力学、信息论、复杂网络理论等. (3)MAS涌现性的控制 MAS涌现性度量的一个重要目的是对MAS 性设计需求,需要进一步的完善和发展. 7 结束语 本文以MAS的涌现性为背景,分析归纳了涌 现的概念和特征,指出了它与总成、自组织等概念的 的涌现行为进行控制,即促使建设性涌现特征出现 和抑制破坏性涌现特征产生.在MAS研究中,有许 多不同的涌现性控制任务.例如,在自组织MAS 中,控制产生自组织涌现特性口。 ;在基于Agent的 区别和联系.考察了MAS面向涌现的宏观描述,归 维普资讯 http://www.cqvip.com 6期 金士尧等:面向涌现的多Agent系统研究及其进展 891 纳出基于逻辑的群体思维涌现性描述和基于统计的 宏观行为涌现性描述.探讨了MAS涌现的微宏观 机制,甄别出受生物、经济和社会等系统启发的机 制.简述了面向涌现的MAS设计方法和设计模式. 通过研究,我们认识到面向涌现的MAS研究取得 了一些阶段性成果,同时存在一些亟待解决的问题, 如涌现性的度量与控制、微宏观机制的动力学等, 在这些方向上有待更进一步的研究. 参 考 文 献 [1] Michael Wooldridge.An Introduction to MuhiAgent Sys— terns.Chichester:John Wiley&Sons,2002(in Chinese) ([英]伍尔德里奇.多Agent系统引论.石纯一,张伟,徐晋晖 等译.北京:电子工业出版社,2()03) [2] Zambonelli F,Omicini A.Challenges and research directions in Agent oriented software engineering.Autonomous Agents and Multi Agent Systems,2004,9(3):253-283 [3] Gabbai J M E,Yin H,Wright W A,Allinson N M.Selbor ganization,emergence and muM—Agent systems//Proceed ings of the IEEE International Conference on Neural Net— works and Brain.Beijing,China,2005:ni124—1863 [4] Holland John H.Hidden Order:How Adaptation Builds Complexity.Redwood City,California:Addison—Wesley Publishing Company,1995(in Chinese) ([美]约翰・H・霍兰.隐秩序:适应性造就复杂性.周晓牧, 韩晖译.上海:上海科技教育出版社,2000) [5] Jochen Fromm.The Emergence of Complexity.Hessen: Kassel University Press,2004 E6] De Wolf T,Holvoet T,Berbers Y.Emergence as a paradigm to engineer distributed autonom Jic software.Department of Computer Science,K.U.Leuven,Belgium:Report CW 380, 2004 [7] Anthony R.Emergence:A paradigm for robust and scalable distributed applications//Proceedings of the 1 st International Conference on Autonomic Computing(ICAC 04).New York,USA,2004:132—139 [8] Palmer D,Kirschenbaum M,Seiter L.Emergence—oriented programming//Proceedings of the International Conference on Systems,Man,and Cybernetics.Hawaii,2005:1441— 1448 [9] De Wolf T. Analysing and Engineering self_organizing emer gent applications[Ph.D.dissertation].Leuven,Belgium: Department of Computer Science 2007 ElO] Goldstein J.Emergence as a construct:History and issues. Emergence,1999,1(1):49—72 [11] De Wolf T,Holvoet T. Emergence versus self_organization: Different concepts but promising when combined//Brueekner S,Di Marzo Serugendo G,Karageorgos A,Nagpal R eds. Engineering Self Organising Systems: Methodologies and Applications.Berlin:Springer—Verlag,2005:1 1 5 [12] Abbott Russ.Emergence explained:Abstractions:Getting epiphenomena to do real work.Complexity,2006,12(1): 13—26 [13] Auyang S Y.Foundations of Complex system Theories:In Economics,Evolutionary Biology,and Statistical Physics. Oxford:Oxford University Press,1998(in Chinese) ([英]欧阳莹之.复杂系统理论基础.上海:上海科技教育出 版社,2002) [14] Ueda K.Synthesis and emergence:Research overview.Arti ficial Intelligence in Engineering,2001,15(4):321—327 [15] Paul Cilliers.Complexity and Postmodernism:Understand— ing Complex Systems.London:Routledge,1998(in Chi— nese) ([南非]保罗・西利亚斯.复杂性和后现代主义.曾国平译. 上海:上海世纪出版集团;上海科技教育出版社,2006) [16] Serugendo G D M,Gleizes M P,Karageorgos A.Self organ isation and emergence in MAS:An Overview.Informatica, 2006,30(1):45 54 [17] Heyligen F. Self_organization,emergence and the architec ture of complexity//Proceedings of the 1st European Confer— ence on System Science.Paris,1989.Paris:AFCET,1992: 23 32 [18] Mamei M,Zambonelli F.Sel ̄Organisation in multi—Agent systems:A middleware approach//Proceedings of the Inter national Workshop on Engineering Self-Organizing Applica tions,Me!bourne,Australia,2003.Berlin=Springer Ver lag,2004:233—248 [19] Mnif M,Mailer Schloer C.Quantitative emergence//Pro— ceedings of the 2006 IEEE Mountain Workshop on Adaptive and Learning Systems(SMCals 2006),Utah State Universi— ty,College of Engineering,Logan,USA,2006:78—84 [2O] Parunak H V D,Brueckner S.Entropy and self_organization in multi—Agent systems//Proceedings of the 5th International Conference on Autonomous Agents.Montreal,Canada, 2001:124—130 [21] Halley J,Winkler D.Classification of emergence and its rela tion to sel organization.Complexity,2008,13(5):10 15 [22] Ulieru Mihaela,Este Robert A.The holonic enterprise and theory emergence:On emergent features of self-organization in distributed virtual Agents.International Journal Cybernet— ics and Human Knowing,2004,11(1):79—98 [23] Dunin—Keplicz Barbara,Rineke Verbrugge.Collective inten— tions.Fundamenta Informaticae,2002,51(3):271—295 [243 Barbara Dunin-Keplicz,Rineke Verbrugge.Collective coln— mitments//Proceedings of the 2nd International Conference on Multi Agent Systems(ICMAS 96).Menlo Park,Califor— nia,1996:56-63 [25] Dunin-Keplicz Barbara,Rineke Verbrugge.Collective moti— vational attitudes in cooperative problem solving//Proceed ings of the 1 st International Workshop of Eastern and Central Europe on Multi-Agent Systems(CEEMAS 99).St.Peters— burg,Russia,1999:22—41 [26] Dunin—Keplicz Barbara.Rineke Verbrugge.Creating tom— mon beliefs in rescue situations//Dunin-Keplicz B,Jankowski 维普资讯 http://www.cqvip.com
892 计 算 机 学 报 2008正 A,Skowron A,Szczuka M eds.Monitoring,Security and iors//Pr0ceed抽gs of the 4th International Joint Conference Rescue Techniques in Multiagent Systems,Series Advances on Autonomous Agents and Multi-Agent Systems in Soft Computing.Berlin:Springer Verlag,2005 69—84 (AAMAS05).Utrecht,Netherlands,2005:305—312 [27] Dunin-Keplicz Barbara,Rineke Verbrugge.A tuning ma— [41] Camponogara Eduardo.On the convergence to and location of chine for cooperative problem solving.Fundamenta lnformat— attractors of uncertain,dynamic games//Bazzan A L C, icae。2004,63(2—3):283—307 Labidi S eds.Proceedings of Advances in Artificial Intelli— [282 Dunin-Keplicz Barbara,Rineke Verbrugge.Calibrating col— gence--SBIA 2004.Berlin:Springer-Verlag,2004:484—493 lective commitments//Marik V,Muller J,Pechoucek M eds. [422 Prokopenko Mikhail,Piraveenan Mahendra Rajah,Wang Pe— Multi—Agent Systems and Applications III.Berlin:Springer- ter.On convergence of dynamic cluster formation in multi— Verlag,2003:73 83 Agent networks//Capcarr ere M S,Freitas A A,Bentley P [29] Michael Wooldridge,Jennings Nicholas R.Towards a theory J,Johnson C G,Timmis J eds.Proceedings of Advances in of cooperative problem solving//Proceedings of the 6th Euro— Artificial Life,8th European Conference in Artificial Life. pean Workshop on Modelling Autonomous Agents:Distribu— Berlin:Springer-Verlag,2005:884—894 ted Software Agents and Applications.Odense,Denmark, [432 Brueckner Sven A,Parunak H Van Dyke.Information-driv— l994.Berlin:Springer Verlag,1996:l5—26 en phase changes in multi—Agent coordination//Proceedings [3O] Paolo Torroni.Computational logic in multi-Agent systems: of Autonomous Agents and Multi—Agent Systems(AAMAS. Recent advanees and future directions.Annals of Mathemat— 2003).Melbourne,Australia,2003:95O一951 ics and Artificial Intelligence,2004,42(1 3):293 305 [442 Foreman Mark,Prokopenko Mikhail,Wang Peter.Phase [31] Tucher Balch.Hierarchic social entropy:An information transitions in self-organising sensor networks//Banzhaf W, theoretic measure of robot group diversity.Autonomous Ro— Christaller T,Dittrich P,Kim J T,Ziegler J eds.Proceed— bots,2000,8(3):209 238 ings of Advances in Artificial Life,the 7th European Confer [32] Prokopenko Mikhail,Wang Peter.Evaluating Team Per— ence on Artificial Life.Berlin:Springer-Verlag,2003:781— formance at the Edge of Chaos.D.Polani et a1.(eds.).Rob— 79l oCup 2003:Robot Soccer World Cup VII.Berlin:Springer- [453 Parunak H Van Dyke,Brueckner Sven,Savit Robert.Uni— Verlag,2004:89-l01 versality in multi Agent systems//Proceedings of the 3rd In— [33] Lerman Kristina,Galstyan Aram.A general methodology ternational Joint Conference on Autonomous Agents and for mathematical analysis of multi-Agent systems.Informa— Multiagent Systems(AAMAS 04).New York,USA,2004: tion Science Institute,University of Southern California: 930—937 Technology Report 529,2001 [46] Holland John H.Emergence:From Chaos to Order.Red— [34] Smolka M,Uhruski P,Schaefer R,Grochowski M.The dy— wood City,California:Addison-Wesley,1998(in Chinese) namics of computing Agent systems//Proceedings of the Workshop on Intelligent Agents in Computing Systems.At— ([美]约翰・霍兰.涌现:从混沌到有序.陈禹等译.上海:上 海世纪出版集团;上海科技教育出版社,2006) lanta,2005:727-734 [35] Johnson Nei1 F,Choe Sehyo C,Gourley Sean,Jarrett Timo— [472 Mano Jean-Pierre,Bourjot Christine,Lopardo Gabriel,Glize thy,Hui Pak Ming.Theory of collective dynamics in multi Pierre.Bio~inspired mechanisms for artificial self-organized Agent complex systems//Kramer B eds.Advances in Solid systems.Informatica,2006,30(1):55 62 State Physics.Berlin:Springer—Verlag,2004,44:427—438 [482 Salima Hassas,Giovanna Di Marzo—Serugendo,Anthony [36] Falcone Rino,Castelfranchi Cristiano.The socio-cognitive Karageorgos,Cristiano Castelfranchi.On self-organising dynamics of trust:Does trust create trust?//Falcone R, mechanisms from social,business and economic domains.In— Singh M,Tan Y—H eds.Trust in Cyber-societies. Berlin: formatica,2006,30(1):63—71 Springer-Verlag,2001:55-72 [49] Lewis Thomas.The Lives of a Cell:Notes of a Biology [37] Liu Hongbo,Abraham Ajith,Clerc Maurice.Chaotic dy— Watcher.New York:Viking Penguin USA,1974(in Chi— namic characte ristics in swarm intelligence.Applied Soft nese) Computing,2007,7(3):lOl9-1026 ([美]刘易斯・托马斯.细胞生命的礼赞.李绍明译.长沙: [38] Lacko Peter,Kvasnicka Vladimir,Pospichal Jiri.An emer— 湖南科学技术出版社,1996) gence of game strategy in multiagent systems.International [503 Brueckner Sven A,Parunak H Van Dyke. Sel organizing Journal of Computational Intelligence and Applications, MANET management//Serugendo,Karageorgos,Rana, 2004,4(3):283—298 Zambonelli eds.Engineering Self-Organizing Applications. [39] Tuyls Karl。Jan T Hoen Pieter Vanschoenwinkel Brain.An Berlin:Springer-Verlag,2004:20—35 evolutionary dynamical analysis of multi—Agent learning in it— [51] Karuna H,Valckenaers P,Saint-Germain B,Verstraete P, erated games.Autonomous Agents and Multi Agent Sys Zamfirescu C B,Van Brussels H.Emergent forecasting terns,2006,l2(1):ll5一l53 using a stigmergy approach in manufacturing coordination and [4o] Parunak H Van Dyke,Brueckner Sven A,Sauter John A, control//Brueckner S et a1.eds.Engineering Self_Organising Matthews Robert.Global convergence of local Agent behav- Systems.Berlin:Springer-Verlag,2005:210—226 维普资讯 http://www.cqvip.com 6期 金士尧等:面向涌现的多Agent系统研究及其进展 893 [52] Luca Tummolini,Cristiano Castelfranchi.Trace signals: [64]Kraus S.Negotiation and cooperation in multi-Agent envi— The meanings of stigmergy//Weyns D,Parunak H V D,Mi ronments.Artificial Intelligent Journal,Special Issue on chel F eds.Environments for Multi Agent Systems III.Ber— Economic Principles of Multi-Agent System,1997,94(I-2): lin:Springer—Verlag,2007:141—156 79—98 [53] Peeters P,Brussel H V,Valckenaers P,Wyns J,Bongaerts [6 5]Sandholm T.Distributed rational decision making//Weiss G L,Kollingbaum M,Heikkila T.Pheromone based emergent ed.Multiagent Systems:A Modern Approach to Distributed shop floor control system for flexible flow shops.Artificial Artificial Intelligence.Cambridge,MA:The MIT Press, Intelligence in Engineering,2001,15(4):343—352 1999:201-258 Es4] Parunak H Van Dyke,Brueckner Sven,Sauter John.Syn [66]Li wan,Tian Sheng-Feng,Huang Hou—Kuan.Evolutionary thetic pheromone mechanisms for coordination of unmanned games theory and Agent self-organizing dynamics.Journal of vehicles//Proceedings of the 1 st International Joint Confer— Computer Research and Development,2006,43(Supple— ence on Autonomous Agents and Multiagent Systems.Bolo— ment):46—50(in Chinese) gna,Italy,2002:448—450 (李万,田盛丰,黄厚宽.进化博弈论及Agent自组织动力学. Ess] Ando Yasushi, Fukazawa Yoshiaki, Masutani Osamu, 计算机研究与发展,2006,43(增刊):46—50) 1wasaki Hirotoshi,Honiden Shinichi.Performance of phero [6 7] Chris Preist,Claudio Bartolini,Andrew Byde.Agent based mone model for predicting traffic congestion//Proceedings of service composition through simultaneous negotiation in for— the 5th Internhtional Joint Conference on Autonomous ward and reverse auctions//Proceedings of the 4th ACM Agents and Multiagent Systems.Hakodate,Japan,2006: Conference on Electronic Commerce(EC一2003).San Diego, 73—8O California,USA,2003:55-63 [56] Morihiro K,Isokawa T,Nishimura H,Matsui N.Emer— [68]wu D J,Sun Yan-Jun.The emergence of trust in multi- gence of flocking behavior based on reinforcement learning// Agent bidding: A computational approach//Proceedings of Proceedings of the 10th International Conference on Knowl— the 34th Hawaii International Conference on System Sciences edge—Based Intelligent Information Engineering Systems. Hawaii,2001.Washington:IEEE Computer Society Press, Bournemouth,England,2006:699-706 2001:1041—1048 [57] Mazurowski Maciej A,Zurada Jacek M.Emergence of com— [69]Fatima Shaheen S,Michael Wooldridge,Jennings Nicholas munication in multi-Agent systems using reinforcement learn— R.An analysis of sequential auctions for common and private ing//Proceedings of the 2006 IEEE International Conference value objects//Proceedings of the 7th International work— on Computational Cybernetics.Talinn,Estonia,2006:1-6 shop on Agent—Mediated Electronic Commerce.Utrecht,The Es8] Dowling Jim,Curran Eoin,Cunningham Raymond,Cahill Netherlands,2005.Berlin:Springer—Verlag,2006:25—38 Vinny.Using feedback in collaborative reinforcement learn— ing to adaptively optimize MANET routing.IEEE Transac— [7O]Voulgaris S,Jelasity M,van Steen M.A robust and scalable Peer-to Peer gossiping protocol//Moro G,Sartori C,Singh tions on Systems,Man,and Cybernetics,Part A,2005,35 M P eds.Agents and Peer to Peer Computing.Berlin: (3):360 372 Springer Verlag,2003:47—58 E59] Hassnaoui A,Bakhouya M,Gaber J.Towards an Agent- based approach for service emergence in pervasive compu [71]Braginsky D,Estrin D.Rumor routing algorithm for sensor ting//Proceedings of the Advanced International Conference networks//Proceedings of the Ist Workshop on Sensor Net on Telecommunications and International Conference on In— works and Applications(WSNA).Atlanta,GA,USA, 2002:22-31 ternet and Web Applications and Services(AICT/ICIW 2006).Guadeloupe,France,2006:15—15 [72]Ramchurn Sarvapali D,Huynh Dong,Jennings Nicholas R. [6o] Ding Yong-Sheng,Sun Hong—Bin,Hao Kuang—Rong.A bio— Trust in multi-Agent systems.Knowledge Engineering Re— inspired emergent system for intelligent Web service composi— view,2004,19(1):1-25 tion and management.Knowledge—Based Systems,2007,20 [73]Cahill V,Gray E,Seigneur J-M,Jensen C D,Chen Yong, (5):457-465 Shand B,Dimmock N,Twigg A,Bacon J,English C,Wa [61] Zhang Wei Ying.Game Theory and Informational Econom— gealla W,Terzis S,Nixon P,Di Marzo Serugendo G,Bryce ics.Shanghai:Shanghai Sanlian Bookstore;Shanghai People s C,Carbone M,Krukow K,Nielson M.Using trust for se— Education Press,1996(in Chinese) cure collaboration in uncertain environments.IEEE Pervasive (张维迎.博弈论与信息经济学.一k海:上海三联书店;上海 Computing Magazine,Special Issue Dealing with Uncertain— 人民出版社,1996) ty,2003,2(3):52—61 [62] Simon Parsons,Michael Woolderidge.Game theory and deci [74]Kamvar S D,Schlosser M T,Garcia-Molina H.The eigen— sion theory in multi-Agent systems.Autonomous Agents and trust algorithm for reputation management in P2P net— Multi-Agent Systems,2002,5(3):243—254 works//Proceedings of the 12th International World Wide [63] Rosenschein J S,Ginsberg M,Genesereth M R.Gooperation Web Conference(WWW2OO3).Budapest,Hungary,2003: without communication//Bond A H,Gasser L eds.Readings 64O一651 in Distributed Artificial Intelligence.San Mateo,California: L75 Di Marzo Serugendo G.Trust as an interaction mechanism Morgan Kaufmann Publishers,1988:220—226 for self_organising systems//Proceedings of the International 维普资讯 http://www.cqvip.com 894 计 算 机 学 报 2008钽 Conference on Complex Systems(ICCS 04).Krakow Po— land,2004:1-8 E76] James Odel1.Agents and emergence.Distributed Compu— ting,1998,10:45—50 [77] De Wolf Tom。Holvoet Tom.Towards a methodology for engineering self-organizing emergent systems//Proceedings of the International Conference on Self-Organization and Ad— aptation of Multi—Agent and Grid Systems(SOAS 2005). Glasgow。Scotland。UK,2005:18—34 [78] Bernon C,Gleizes M—P,Peyruqueou S,Picard G.ADELFE: A methodology for adaptive multi—Agent systems engineer— ing//Proceedings of the 3rd International Workshop on Engi— neering Societies in the Agents World(ESAW一2002).Ma— drid,Spain,2002.Berlin:Springer-Verlag,2003:156-169 [79] Fiona Polack,Susan Stepney.Emergent properties do not re— fine.Electronic Notes in Theoretical Computer Science, 2005,137(2):163-181 [8O] Poulton G,Guo Y,Valencia P,James G,Prokopenko M, Wang P.Designing enzymes in a multi—Agent system based on a genetic algorithm//Groen F et al eds.Intelligent Auton— omous Systems 8(IAS一8).Amsterdam:IOS Press,2004: 253—262 [81] Poulton Geoff,Guo Ying,James Oeoff,Valencia Phil, Gerasimov Vadim,Li Jia-Ming.Directed self-asembly of 2一 Dimensional mesoblocks using top—down/bottom—up design// Brueckner S et al eds.Engineering Self-Organizing Systems: Methodologies and Applications.Berlin:Springer-Verlag, 2005:154—166 [82] De Wolf Tom,Holvoet Tom,Samaey Giovanni.Engineering se1f_organising emergent systems with simulation-based sci— entific analysis//Brueckner S,Di Marzo Serugendo G,Hales D,Zambonelli F eds.Proceedings of the 3rd International Workshop on Engineering Se1 0rganising Applications. Utrecht,The Netherlands:Universiteit Utrecht,2005:146— 16O [83] Gardelli Luca,Viroli Mirko,Omicini Andrea.On the role of simulation in the engineering of self-organizing systems:De— tecting abnormal behaviour in MAS//Proceedings of the Workshop on Objects and Agents(W0A O5).Camerino。 JIN Shi—Yao,born in 1937,profes— sor,Ph.D.supervisor.His current re— search interests include systems model— ing and simulation,multi-agent sys— tems,distributed computing,and virtual teahzation. Background .In a lot of applications,MASs are identified as CASs. So,in the area of MAS.more and more attentions are paid to the phenomenon of emergence which is intrinsic in CASs. Mc,Italy,2005:85—90 [84] Gamma E,Helm R,Johnson R,Vlissides J.Design Pat— terns:Elements of Reusable Object-0riented Software.Red— wood City,California:Addison Wesley Professional,1995 (in Chinese) (Gamma E,Helm R,Johnson R,Vlissides J.设计模式:可 复用面向对象软件的基础.李英军,马晓星,蔡敏,刘建中等 译.北京:机械工业出版社,2000) [85] De Wolf Tom。Holvoet Tom.Design patterns for decentral— ized coordination in self-organizing emergent systems// Brueckner S,Hassas S,Jelasity M,Yamins D eds.Engi— neering Self-Organizing Systems.Berlin:Springer-Verlag, 2007:28-49 [86] De Wolf T,Holvoet T.A catalogue of decentralized coordi— nation mechanisms for designing self_organizing emergent ap— plications.Department of Computer Science.K.U.Leuven, Belgium:Report CW 458,2006 [87] Gardelli Luca,Viroli Mirko,Omicini Andrea.Design pat— terns for self-organizing multivalent systems//Proceedings of the International Workshop on Engineering Emergence in De— centralised Autonomic Systems(EEDAS 2007),ICAC 2007, Jacksonville,FL,USA,2007.University of Greenwich: CMS Press,2007:61-70 [88] Lind J.Patterns in Agent—oriented software engineering// Proceedings of the 3rd International Workshop on Agent OA— ented Software Engineering(A0SE 2002).Bologna,Italy, 2002.Berlin:Springer-Verlag,2003:47—58 [89] Rao A S,Georgeff M P.Modelling rational Agents within a BDI—architecture//James Alien,Richard Fikes,Erik Sand— ewall eds.Proceedings of the 2nd International Conference of Principles of Knowledge Representation and Reasoning. San Mateo,California:Morgan Kaufmann Publishers,1991: 473—484 [9O] Brafman Ronen I,Tennenholtz Moshe.Modeling Agents as qualitative decision makers.Artificial Intelligence,1997,94 (1-2):217-268 [91] Kephart J.Software Agents and the route to the information economy.Proceedings of the National Academy of Sciences of the United States of America,2002,99(3):7207—7213 HUANG Hong—Bing,born in 1977,Ph.D.candidate. His current research interests include multi—agent systems, complex systems modeling and simulation. FAN Gao-Jun,born in 1980,P h_D.candidate.His current research interests include multi—agent systems,com— plex systems modeling and simulation. And emergence—oriented research on MASs becomes a new direction in the area of MAS.Emergence—oriented researches focus on the emergent characteristics of MASs and the mech— 维普资讯 http://www.cqvip.com 6期 金士尧等:面向涌现的多Agent系统研究及其进展 895 anisms of emergence which correlate the macro—and micro— In this paper,after an introduction of the concept of level characteristics.The final goal of these researches is to emergence,the emergence—oriented approaches for character— develop a set of methods for engineering emergence in izing emergent behaviors of MASs are reviewed.Then the MASs. authors give an overview of the micro-macro mechanisms of In this area,the concept of emergence has been debated emergence in MASs,and a survey of emergence—oriented en— for a long time,and is not settled yet.To study the emergent gineering methodologies and design patterns of MASs.Final— characteristics。the laws governing the evolution of MASs ly,the problems of these researches are discussed;and fu— must be fined.This leads to the emergence—oriented descrip— ture directions in this area are given by the authors. tion of MAS.At present,there are mainly two approaches: Authors have done research on MAS and CAS,especial— Logic—based and statistics—based.To engineer emergence in ly on MAS-based modeling and simulation of CASs,for many MASs at runtime,the proper way is tO adopt relevant mecha— years.By these studies,they have a deep understanding of nisms for emergence and to introduce appropriate artifact for emergence in CASs.Recently,they apply their research re— controlling.So design patterns using emergent mechanisms sult to emergence—oriented research on MASs.A remarkable are very popular.Most of the mechanisms,nOW,are come tool,namely JCass,have facilitated their research,which is form biology,economic and society.And the engineering a distributed platform for Agent-based simulations developed methodologies are extends of the traditional software engi— by themselves. neering methodologies.
因篇幅问题不能全部显示,请点此查看更多更全内容