Accession Number : ADA454661


Title :   Advanced Agent Methods in Adversarial Environment


Descriptive Note : Final rept. 14 Jun 2004-14 Dec 2005


Corporate Author : CZECH TECHNICAL UNIV PRAGUE


Personal Author(s) : Pechoucek, Michal ; Rehak, Martin ; Sislak, David ; Benda, Petr ; Foltyn, Lukas ; Tozicka, Jan ; Jisl, Pavel ; Volf, Premysl


Full Text : http://www.dtic.mil/get-tr-doc/pdf?AD=ADA454661


Report Date : 30 NOV 2005


Pagination or Media Count : 135


Abstract : The suggested research effort aims at continuation of the research in the field of social knowledge, social behavior and meta-reasoning in multi-agent systems. The main and fundamental theme of this research project is extending meta-reasoning and social knowledge manipulation to the problem of intent modeling in the non-trivial domains of non-cooperative, competitive, i.e. adversarial behavior. The main ambition is to extend meta-reasoning for the problem of intent modeling in the non-cooperative, cooperative, i.e. primarily adversarial domain. This effort retains the focus on environments with partial communication inaccessibility. This extension of the metareasoning to adversarial agent behavior is a natural step in applicability of this technology to Information Operations. Dr. Pechoucek has a long and successful history with AFRL working under grants to Czech Tech University with Dr. Vladimir Marik. Deliverable 1: (month 6) Interim report defining the concept of non-collaboratively and hostility in multi-agent systems. with design of a scenario for investigating agents social behaviour in non-collaborative and adversarial environment. Deliverable 2: (month 12) Interim report describing potential role of agent's acquaintance model, social knowledge and meta-reasoning in non-collaborative environment with partial communication accessibility. Deliverable 3: (month 18) Comprehensive final report providing technical description of the entire research effort including formal definition of measures and quantities of agents collective behaviour, adapted social knowledge model, meta-reasoning model and coalition formation algorithms for non-collaborative and adversarial environments and proof-of-concept experimental verification.


Descriptors :   *REASONING , *ADVERSE CONDITIONS , COMPETITION , ARTIFICIAL INTELLIGENCE , COMMAND AND CONTROL SYSTEMS


Subject Categories : CYBERNETICS
      PSYCHOLOGY
      COMMAND, CONTROL AND COMMUNICATIONS SYSTEMS


Distribution Statement : APPROVED FOR PUBLIC RELEASE