Emergent Intelligence of Networked Agents

Recently,the study of intelligenceemerged from interactionsamong many agentshasbeenpopular. Inthisstudyitisrecognizedthatanetworkstructure oftheagentsplaysanimportantrole. Thecurrentstate-of-theartinage- based modeling tends to be a mass of agents that have a series of states thattheycanexpressasaresultofthenetworkstructureinwhichtheyare embedded. Agentinteractionsofallkindsareusuallystructuredwithcomplex networks. Researchoncomplexnetworksfocusesonscale-freenessofvarious kindofnetworks. Computationalmodelingofdynamicagentinteractionsonrichlystr- turednetworksisimportantforunderstandingthesometimescounter-intuitive dynamicsofsuchlooselycoupledsystemsofinteractions. Yetourtoolsto model,understand,andpredictdynamicagentinteractionsandtheirbe- vioroncomplexnetworkshavelaggedfarbehind. Evenrecentprogressin networkmodelinghasnotyeto?eredusanycapabilitytomodeldynamic processesamongagentswhointeractatallscalesonsuchassmall-worldand scale-freenetworks. Generallythehigh-dimensional,non-linearnatureofthe resultingnetwork-centricmulti-agentsystemsmakesthemdi?cultorimp- sibletoanalyzeusingtraditionalmethods. Agentsfollowlocalrulesunder complexnetworkconstraints. Theideaofcombiningmulti-agentsystemsand complexnetworksisalsoparticularlyrichandfreshtofostertheresearchon thestudyofverylarge-scalemulti-agentsystems. Weintendtoturnthisintoanengineeringmethodologytodesigncomplex agentnetworks. Multi-agentnetworkdynamicsinvolvesthestudyofmany agents,constituentcomponentsgenerallyactiveoneswithasimplestructures andwhosebehaviorisassumedtofollowlocalrules,andtheirinteractionson complexnetwork. Abasicmethodologyistospecifyhowtheagentsinteract, andthenobserveemergentintelligencethatoccuratthecollectivelevelin ordertodiscoverbasicprinciplesandkeymechanismsforunderstandingand shapingtheresultingintelligentbehavioronnetworkdynamics. Thevolumecontainsrefereedpapersaddressingvariousimportanttopics thataimsattheinvestigationofemergentintelligenceonnetworkedagents. vi Preface Especiallymostpapershighlightonthetopicssuch"networkformationamong agents", "in?uenceofnetworkstructuresonagents", "network-basedcoll- tivephenomenaandemergentintelligenceonnetworkedagents". TheselectedpapersofthisvolumewerepresentedattheWorkshopon EmergentIntelligenceofNetworkedAgents(WEIN06)attheFifthInt- nationalJointConferenceonAutonomousAgentsandMulti-agentSystems (AAMAS2006),whichwasheldatFutureUniversity,Hakodate,Japan,from May8to12,2006. WEIN06isconcernedwithemergenceofintelligentbe- viorsovernetworkedagents andfosteringtheformationofanactivemul- disciplinarycommunityonmulti-agentsystemsandcomplexnetworks. We especiallyintendedtoincreasetheawarenessofresearchersinthesetwo?elds sharingthecommonviewoncombiningagent-basedmodelingandcomplex networksinordertodevelopinsightandfosterpredictivemethodologiesin studyingemergentintelligenceonofnetworkedagents. Fromthebroadsp- trumofactivities,leadingexpertspresentedimportantpaperandnumerous practicalproblemsappearthroughoutthisbook. Weinvitedhighqualityc- tributionsonawidevarietyoftopicsrelevanttothewideresearchareasof multi-agentnetworkdynamics. Weespeciallycoveredin-depthofimportant areas including: Adaptation and evolution in complex networks, Economic agentsandcomplexnetworks,Emergenceincomplexnetworks,Emergent- telligenceinmulti-agentsystems,Collectiveintelligence,Learningandevo- tioninmulti-agentsystems,Webdynamicsascomplexnetworks,Multi-agent basedsupplynetworks,Network-centricagentsystems,Scalabilityinmul- agentsystems,Scale-freenetworks,Small-worldnetworks.