� � � Emergent Properties for Data Distribution in � a Cognitive MAS � ——————————————————————— � � Andrei Olaru, Cristian Gratie, Adina Magda Florea University ’Politehnica’ of Bucharest � � 10.2009 � � 1/ 15 Computer . . Andrei Olaru, Cristian Gratie, Adina Magda Florea Science . IDC 2009 & Engineering . Aya Napa, Cyprus, 10.2009 Department
� Introduction � Definitions � Reactive Systems � Cognitive Agents Emergent Properties for Data � Cognitive Emergence Distribution in a Cognitive MAS � Design ——————————————- � Agents overview � Goals � Example � Results � Conclusion 2/ 15 Computer . . Andrei Olaru, Cristian Gratie, Adina Magda Florea Science . IDC 2009 & Engineering . Aya Napa, Cyprus, 10.2009 Department
Emergence – essential issue in the engineering of multi-agent � systems. � Emergent Properties for Data � Distribution in a Cognitive MAS · lower (micro) level – simple entities that interact. � Introduction · higher (macro) level – complex behaviour of the system as � Definitions a whole. � Reactive Systems · Most studies of emergence use reactive � Cognitive Agents agent systems. � Cognitive Emergence · Cognitive systems are more capable. � Design · Emergents in cognitive agent systems � Agents may be more complex and, therefore, � Goals more useful. � Example Purpose: Create a multi-agent system � Results for the storage and distribution of data, formed of cognitive agents, designed so � Conclusion that it would manifest the said emergent behaviour. 3/ 15 Computer . . Andrei Olaru, Cristian Gratie, Adina Magda Florea Science . IDC 2009 & Engineering . Aya Napa, Cyprus, 10.2009 Department
� � Emergent Properties for Data Emergence is: � Distribution in a Cognitive MAS · the concept of some new phenomenon arising in a system � Introduction that wasn’t in the system’s specification to start with. � Definitions of Emergence [Standish, 2001] · coherent emergents at the macro-level that � Reactive Systems dynamically arise from the interactions between the � Cognitive Agents parts at the micro-level. Such emergents are novel with respect to the individual parts of the system. � Cognitive Emergence [De Wolf and Holvoet, 2005] � Design · in the context of an interacting set of agents whose dynamics � Agents are expressed in a vocabulary D , a global phenomenon – static or dynamic, but nevertheless invariant – that is observed � Goals by the agents or by an external observer and can only be � Example interpreted in a vocabulary D ′ that is different from D . � Results [Beurier et al., 2002] � Conclusion Important: emergence allows obtaining a function / behaviour of higher level from the interaction of lower level entities. 4/ 15 Computer . . Andrei Olaru, Cristian Gratie, Adina Magda Florea Science . IDC 2009 & Engineering . Aya Napa, Cyprus, 10.2009 Department
� � � · Reactive agents – simple behaviour: � � interact � Emergent Properties for Data act � Distribution in a Cognitive MAS � Introduction change perceive � Definitions environment � Emergence in Reactive Agent Systems � Cognitive Agents � Cognitive Emergence � Design � Agents [Beurier et al., 2002] [Zambonelli et al., 2004] � Goals � Example � Results � Conclusion [Picard and Toulouse, 2005] · emergents are of the same nature as the properties of individuals, but at a higher level 5/ 15 Computer . . Andrei Olaru, Cristian Gratie, Adina Magda Florea Science . IDC 2009 & Engineering . Aya Napa, Cyprus, 10.2009 Department
� � � � Cognitive agents feature: � Emergent Properties for Data � Distribution in a Cognitive MAS information about self ◮ beliefs / knowledge � Introduction ◮ what it wants to do ◮ desires / goals ◮ what it is able to do � Definitions ◮ intentions ◮ how it can do it � Reactive Systems ◮ plans � Cognitive Agents Components: � Cognitive Emergence � Reasoner input � Design � Agents Planner � Goals � Example output � Results Scheduler � Conclusion 6/ 15 Computer . . Andrei Olaru, Cristian Gratie, Adina Magda Florea Science . IDC 2009 & Engineering . Aya Napa, Cyprus, 10.2009 Department
� � � � � � Expected behaviour: � Emergent Properties for Data � Distribution in a Cognitive MAS higher level global goal � Introduction � Definitions emergence local goals and plans � Reactive Systems · do not contain global goal lower level � Cognitive Agents · plans do not explicitly lead do global goal � Emergence in Cognitive Agent Systems � Design � Agents � Goals · Cognitive emergence – achieving a high level global goal � Example through the interaction between agents that follow their own, � Results individual, possibly selfish, goals. � Conclusion 7/ 15 Computer . . Andrei Olaru, Cristian Gratie, Adina Magda Florea Science . IDC 2009 & Engineering . Aya Napa, Cyprus, 10.2009 Department
· Design a multi-agent system in which neighbour (or � � Emergent Properties for Data acquainted) agents exchange information based on local goals � Distribution in a Cognitive MAS (with no centralised control), so that, globally: � Introduction � Definitions · when an agent produces a piece of information (by user input or aggregation), eventually the information is known by the � Reactive Systems agent(s) that might be interested in it. � Cognitive Agents · when an agent needs a certain piece of information, � Cognitive Emergence eventually it will come to know it. � System design System specifications: � Agents � Goals ◮ Cognitive agents placed in a rectangular grid. � Example ◮ Agents communicate directly only with their 8 neighbours. � Results ◮ Agents have a limited storage of information. � Conclusion 8/ 15 Computer . . Andrei Olaru, Cristian Gratie, Adina Magda Florea Science . IDC 2009 & Engineering . Aya Napa, Cyprus, 10.2009 Department
� � � � � · BDI-inspired model � � Emergent Properties for Data Beliefs: � AgentID , DataID � � Distribution in a Cognitive MAS example: � B , D 1 � = agent B has data D 1 � Introduction Goals: � Goal , DataID � � Definitions example: � Get , D 3 � = must get data D 3 � Reactive Systems Actions: � Action , related agent and belief or goal � � Cognitive Agents example: � Inform , C , � B , D 1 �� = inform agent C that agent B has data D 1 � Cognitive Emergence agent behaviour � Design � Planner � Scheduler Reasoner � Agent design � Goals assimilate � Example make a plan execute plan new knowledge � � � � � � � Results � � � � � � � � � � � � � � � � Conclusion � � � � � � knowledge base ongoing plans waiting plans 9/ 15 Computer . . Andrei Olaru, Cristian Gratie, Adina Magda Florea Science . IDC 2009 & Engineering . Aya Napa, Cyprus, 10.2009 Department
· the goals of an agent are: � � Emergent Properties for Data ◮ In case there is an external request for data, provide that data or � Distribution in a Cognitive MAS try to find it in the vicinity. � Introduction ◮ Maintain 25% of the capacity free, ready for potential data � Definitions coming from the environment. � Reactive Systems ◮ In case there is available capacity (over 25%), request interesting data from a neighbour. � Cognitive Agents ◮ Process and respond to messages from neighbour agents. � Cognitive Emergence ◮ If all other objectives are complete (capacity 75%, no messages), discard some data that is already contained by most of the � Design surrounding agents. � Agents · available actions: � Agent goals ◮ Send a request for data to a neighbour. � Example ◮ Send a piece of data to a neighbour (as reply to a request only). ◮ Receive data or knowledge from a neighbour. � Results ◮ Discard a piece of data. � Conclusion ◮ Broadcast to the all neighbours own knowledge and intentions. 10/ 15 Computer . . Andrei Olaru, Cristian Gratie, Adina Magda Florea Science . IDC 2009 & Engineering . Aya Napa, Cyprus, 10.2009 Department
� � Emergent Properties for Data · example behaviour, over one step: � Distribution in a Cognitive MAS � Introduction � Definitions � Reactive Systems � Cognitive Agents � Cognitive Emergence � Design � Agents � Goals � Example � Results � Conclusion 11/ 15 Computer . . Andrei Olaru, Cristian Gratie, Adina Magda Florea Science . IDC 2009 & Engineering . Aya Napa, Cyprus, 10.2009 Department
� � Emergent Properties for Data · example behaviour, over one step: � Distribution in a Cognitive MAS � Introduction � Definitions � Reactive Systems � Cognitive Agents � Cognitive Emergence � Design � Agents � Goals � Example � Results � Conclusion 11/ 15 Computer . . Andrei Olaru, Cristian Gratie, Adina Magda Florea Science . IDC 2009 & Engineering . Aya Napa, Cyprus, 10.2009 Department
Recommend
More recommend
