P.J. Gmytrasiewicz and E.H. Durfe. A rigorous, operational formalization of recursive modeling. In Proceedings of the First International Conference on Multi-Agent Systems (ICMAS), pp. 125--132, 1995.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....and when an agent modifies its reactive component, it should broadcast information about this modification to the other agents. In this way we are still able to make linear anticipations. This approach can be contrasted with the Recursive Modeling Method suggested by Gmytrasiewicz and Durfee [4] in which an agent modeling another agent includes that agent s models of other agents and so on, resulting in a recursive nesting of models. 4.2 Linearly Anticipatory Agents versus A based Agents for Path Finding in More Complex Worlds In this experiment, which is described by Seger and ....

P.J. Gmytrasiewicz and E.H. Durfee. A rigorous, operational formalization of recursive modeling. First International Conference on Multiagent Systems, pages 125-132. AAAI Press, 1995.

.... non cooperative game theory (which is particularly concerned with providing equilibrium strategies in which no agent wants to change its strategy whatever its opponents do) is an important approach for analysing strategic interactions among agents [79] 148] The recursive modelling method [154] [49] is employed by an agent to reason about its opponent so that it can generate its own strategy in response. In [171] a Bayesian network is used to update the knowledge and belief that each agent has about the environment and other agents; and offers and counter offers between agents are generated ....

P. Gmytrasiewicz and E. Durfee. A rigorous, operational formalization of recursive modeling. In Proceedings of the First International Conference on Multi-Agent Systems, pages 125 132, 1995.

....That is, different auctions yield different expected revenue. Our method can be used to design better auction protocols, given the participants belief structures. 1. Introduction Reasoning about others and interactive knowledge have been the subject of continuous interest in multiagent systems [11,12,13,20], artificial intelligence [6,7,8] and game theory [1,3,15] In multiagent interaction, where an agent s action interferes with other agents actions, hierarchies of beliefs arise in an essential way. Usually an agent s optimal decision depends on what he believes the other agents will do, which in ....

....such beliefs, agents need some finite and computationally tractable way to represent them. The second issue that deserves consideration is the feasibility of decision making based on infinitely nested beliefs. Finite hierarchies of beliefs have been studied by Gmytrasiewicz, Durfee and Vidal [11,12,13, 20]. The main advantage of their recursive modeling method is that a solution can always be derived. The recursive modeling method is based on the assumption that once an agent has run out of information his belief hierarchy can be cut at the point where there is no sufficient information. At the ....

[Article contains additional citation context not shown here]

Gmytrasiewicz P., Durfee E. A Rigorous, Operational Formalization of Recursive Modeling. In Proceedings of ICMAS'95, pp. 125-132, 1995.

....and intentions (namely B h B s , B h I s , I h B s , and I h I s ) and the contexts for the hearer s beliefs and intentions regarding the speaker s beliefs and intentions regarding the hearer s beliefs and intentions. Of course, this nesting could be extended inde nitely (for more details see [11, 12, 16]) but three levels (as depicted in gure 1, where circles represent contexts and arrows represent bridge rules) are sucient to illustrate the abductive methods for the inference of mental states from communicative acts. 2.2 Mental States The logical systems that formalize the reasoning with a ....

P.J. Gmytrasiewicz and E.H. Durfe. A rigorous, operational formalization of recursive modeling. In Proceedings of the First International Conference on Multi-Agent Systems (ICMAS), pages 125-132, 1995.

....bidding strategies for agents participating in online auctions. The Recursive Modeling Method (RMM) uses a decision theoretic paradigm of rationality, where an agent makes decisions based on what they think the other agents are likely to do, and what the other agents think about them and so on [Gmytrasiewicz and Durfee, 1995]. The downside to this approach is that not all the information in the recursive model may be relevant to the agent; there may be cases where this information does not influence the agent s decision making at all. It is also possible that little or no information is available for the agent to use ....

P. J. Gmytrasiewicz and E. H. Durfee. A rigorous, operational formalization of recursive modeling. In Proceedings of the First International Conference on Multi-Agent Systems (ICMAS), Menlo Park, California, 1995, pp. 125-132.

....and intentions (namely B h B s , B h I s , I h B s , and I h I s ) and the contexts for the hearer s beliefs and intentions regarding the speaker s beliefs and intentions regarding the hearer s beliefs and intentions. Of course, this nesting could be extended indefinitely (for more details see [9, 10, 14]) but three levels (as depicted in figure 1, where circles represent contexts and arrows represent bridge rules) are sufficient to illustrate the abductive methods for the inference of mental states from communicative acts. I h s B h I I h I s B h B h B s B h I s B h B s h B B h B s I h h B ....

P.J. Gmytrasiewicz and E.H. Durfe. A rigorous, operational formalization of recursive modeling. In Proceedings of the First International Conference on Multi-Agent Systems (ICMAS), pages 125--132, 1995.

....expert than an intermediate opponent. The expert opponent is expected to maximize its utility, making few mistakes, and the intermediate opponent is not expected to perceive all of its options, making more mistakes. Identifying another agent s model is crucial in effectively playing against it (Gmytrasiewicz and Durfee, 1995; Tambe and Gmytrasiewicz, 1996) How does an agent determine the skill level of an unknown opponent, or how does it determine which model applies The agent can assume that the opponent utilizes the same behavioral strategy as itself. As the opponent 22 takes moves that are not optimal as ....

Gmytrasiewicz, P. J. and Durfee, E. H. (1995). A rigorous, operational formalization of recursive modeling. In Lesser, V., editor, Proceedings of the First International Conference on Multi--Agent Systems, pages 125--132, San Francisco, CA. MIT Press.

....is (im)possible. This is one area where microeconomics and computer science fruitfully blend. Another area of substantial current and potential future cross fertilization is the relaxation of the common knowledge assumption that underlies the Nash equilibrium solution concept and its refinements [14, 18]. In the future, systems will increasingly be designed, built, and operated in a distributed manner. A larger number of systems will be used by multiple real world parties. The problem of coordinating these parties and avoiding manipulation cannot be tackled by technological or economic methods ....

P. J. Gmytrasiewicz and E. H. Durfee. A rigorous, operational formalization of recursive modeling. In Proceedings of the First International Conference on Multi-Agent Systems (ICMAS-95), pages 125--132, San Francisco, CA, June 1995.

....favorable results, weakened if unfavorable. The most important choice for an agent within an adaptive multiagent system is whether to model the other agents in the system and compute optimal actions based on this model and knowledge of the reward structure of the game (model based learning) (Gmytrasiewicz Durfee 1995), or to directly learn the expected utility of actions in a given state (direct learning) A direct learning approach has been proposed for sequential games, where agents increase the probability of playing actions that have met with success in previous periods. This is a version of Qlearning, ....

Gmytrasiewicz, P. J., and Durfee, E. H. 1995. A rigorous, operational formalization of recursive modeling.

....of the calendar of another agent. An important distinction here is that they are modeling primarily the internal state of the other agent than the decision policy being used by that agent. Gmytrasiewicz and Durfee presents a decision theoretic approach to updating recursive models of other agents (Gmytrasiewicz Durfee 1995). Their model updating procedure, however, is based more on assumptions of rationality of the other agent and is not dependent on observed behavior of other agents. Zeng and Sycara present a learning mechanism by which agents can learn about payoff structures of other agents in a sequential ....

Gmytrasiewicz, P. J., and Durfee, E. H. 1995. A rigorous, operational formalization of recursive modeling.

....based on the utilities resulting from those actions. A problem in multiagent systems is that the best action for Agent A i might be in conflict with that for another Agent A j . Agent A i , then, should try to model the behavior of A j , and incorporate that into its expected utility calculations (Gmytrasiewicz Durfee 1995). The optimal action for an individual agent might not be the optimal action for its group. Thus an agent can evaluate the utility of its actions on two levels: individual and group. The group level calculations require more information and impose greater cognitive load, whereas the individual ....

Gmytrasiewicz, P. J., and Durfee, E. H. 1995. A rigorous, operational formalization of recursive modeling.

....contenders competed for developing optimized trading strategies. However, the main concern of our proposal consists in providing a method for performing multi agent reasoning under uncertainty based on the modelling of the other agents behaviour likewise [18] where the recursive modelling method [7] was used for constructing agents capable of predicting the other agents behaviour in Double auction markets. At present, a proof of concept implementation of our proposal is undergoing empirical evaluation. We are basically analyzing which utility and similarity functions yield good ....

P. Gmytrasiewicz and E. H. Durfee. A rigorous, operational formalization of recursive modeling. In Proceedings of the First International Conference on Multi-Agent Systems, pages 125--132, 1995.

....of reasoning is defined as a transformation that conveys the knowledge from higher types, in the hierarchy, into the lower types and finally into the ground type. The final ground knowledge is the basis for determining the final intention. Similar ideas of knowledge transformations may be found in [7], 13] 14] where a special kind of agent rationality is considered, namely Bayesian rationality. However, the idea of the ground type is not distinguished explicitly there. Moreover, in all the above papers only, a so called, static case is considered, that is, agents take actions only once. The ....

....0 1 g. The rest components of the mutual knowledge are empty sets and denote that agent i has no knowledge about the agent j 00 1 , and no knowledge about what agents j 1 ; j 0 1 know about other agents. Similar representation of mutual knowledge is applied in Recursive Modeling Method, see [7]. There is also another representation of mutual knowledge that is much more simple to grasp, however is hard to use in applications. It may be called generic representation, and is constructed in the following way. Let K i denote the type of agent i s knowledge, that will be defined below. ....

P. J. Gmytrasiewicz, and E. H. Durfee.: A rigorous, operational formalization of recursive modeling. In Proc. First International Conference on Multiagent Systems. San Francisco, California (1995) 125--132.

....plan recognition task in the air combat simulation environment, while [2] explored the use of finite automata to model the opponent agent s strategy. A series of papers reported works on recursive modeling method (RMM) for decision theoretic agents, which uses deeper, nested models of other agents [7, 15, 5, 4, 12]. RMM represents an agent s decision situation in the form of a payoff matrix. In terms of belief, desire and intention (BDI) architecture, a payoff matrix contains a compiled representation of the agent s capabilities, preferences, and beliefs about the world. Beliefs about other agents are ....

P. J. Gmytrasiewicz and E. H. Durfee. A rigorous, operational formalization of recursive modeling. In Proceedings of the First International Conference on Multi-Agent Systems (ICMAS), pages 125--132, June 1995.

....making rational decisions. The Recursive Modeling Method (RMM) draws on game theory for its inspiration, but its motivations are those of artificial intelligence: to provide reasoning mechanisms for computational agents (players in the game) that make rational decisions in multiagent situations (Gmytrasiewicz Durfee, 1995). Rationality, for RMM, is of the decisiontheoretic kind: an agent should make a decision that maximizes its expected utility, given everything that it knows about the situation (including the other agents) Agents make decisions individually and independently, but to the extent that the knowledge ....

Gmytrasiewicz, P.J., and Durfee, E. H. 1995. A Rigorous, Operational Formalization of Recursive Modeling. In Proceedings of the First International Conference on Multi-Agent Systems, pages 125-132, San Francisco, CA.

....system wide properties of markets populated with self interested agents. To address the question of what an agent should do in multiagent interactions, some researchers have developed a new solution concept based on a recursive model of other agents (i.e. what they think what I think and so on) [7]. While game theoretic agents are ultra smart and super rational, such that they can reason about this whole recursive hierarchy ad infinitum [2] the recursive modeling method (RMM) assumes that agents can only build a finite nesting of models (because of practical limitations on acquiring such ....

Gmytrasiewicz, P.J. and E.H. Durfee. A Rigorous, Operational Formalization of Recursive Modeling. in First International Conference on Multi-Agent Systems (ICMAS-95). 1995. 125-132.

....Markovitch (1996) explored the use of finite automata to model the opponent agent s strategy. A series of papers reported works on recursive modeling method (RMM) for decision theoretic agents, which uses deeper, nested models of other agents (Gmytrasiewicz et al. 1991, Vidal and Durfee, 1995, Gmytrasiewicz and Durfee, 1995, Gmytrasiewicz, 1996, Noh and Gmytrasiewicz, 1997) RMM represents an agent s decision situation in the form of a payoff matrix. In terms of belief, desire and intention (BDI) architecture, a payoff matrix contains a compiled representation of the agent s capabilities, preferences, and beliefs ....

Gmytrasiewicz, P. J., and Durfee, E. H. (1995). A rigorous, operational formalization of recursive modeling. In Proceedings of the First International Conference on Multi-Agent Systems (ICMAS), 125--132.

....relying on protocols computed by a designer beforehand could lock the agents into suboptimal behavior in unpredictable domain like the battlefield, in which situations that were not foreseen by the designer are likely to occur. Our approach uses the Recursive Modeling Method proposed before in [4]. RMM endows an agent with a compact specialized representation of other agents beliefs, abilities, and intentions. As such, it allows the agent to predict the message s decision theoretic (DT) pragmatics, i.e. how a particular message will change the decision making situation of the agents, and ....

....like the one in Figure 6, can be solved using dynamic programming. The solution proceeds bottom up and results in expected utility values being assigned to Battery2 s alternative behaviors, the best of which can then be executed. For a more detailed solution procedure and further discussion, see [4, 12]. 3 Communication in Anti Air Defense We identify a communicative act with its decision theoretic (DT) pragmatics, defined as the transformation of the state of knowledge about the decision making situation (i.e. the recursive model structure) the act brings about. We model DT pragmatics using ....

P. J. Gmytrasiewicz and E. H. Durfee. A rigorous, operational formalization of recursive modeling. In Proceedings of the First International Conference on Multi-Agent Systems, pages 125--132, Menlo Park, 1995. AAAI Press/The MIT Press.

....and penalties, but computes them based on possibly incomplete and uncertain models of other agents. Tambe s work is further closely related to prior work by Cohen [2] and Grosz [8] As we mentioned, we investigate decision theoretic message selection using the Recursive Modeling Method (RMM) [6, 4]. RMM endows an agent with a representa tion of other agents beliefs, abilities, and intentions. As such, it allows the agent to predict the message s pragmatics, i.e. how a particular message will change the state of knowledge of the other agent(s) and how the other agents are likely to ....

P. J. Gmytrasiewicz and E. H. Durfee. A rigorous, operational formalization of recursive modeling. In Proceedings of the First International Conference on Multi-Agent Systems, pages 125--132, Menlo Park, 1995. AAAI Press/The MIT Press.

....the hostile missiles. Based on these attributes combined, each unit has to determine the optimal action from his probabilistic decision model. For the purpose of coordinated decision making in a multiagent environment, our research uses the Recursive Modeling Method (RMM) previously reported in [2, 3]. RMM enables an agent to model the other agents and to rationally coordinate with them even if no protocol or overall plan can be established explicitly in advance. Using RMM as a decision making tool, an agent rationally selects his action under uncertainty guided by the principle of expected ....

....by independent defense units. 3 Decision Theoretic Agent To be rational in decision theoretic sense, the agents follow the principle of maximum expected utility (PMEU) 10] In this section, we will show how PMEU can be implemented in this case study using the Recursive Modeling Method (RMM) RMM [2, 3] will be used to model the other agent, and to select the most appropriate missile to intercept by a given defense battery. 3.1 An Example Scenario Our approach is to take the agent oriented perspective. In the examples scenario (Fig. 1) we view the decision making through the eyes of an ....

P. J. Gmytrasiewicz and E. H. Durfee. A rigorous, operational formalization of recursive modeling. In Proceedings of the First International Conference on MultiAgent Systems, pages 125--132. AAAI Press/The MIT Press, 1995.

....relying on protocols computed by a designer beforehand could lock the agents into suboptimal behavior in unpredictable domain like the battlefield, in which situations that were not foreseen by the designer are likely to occur. Our approach uses the Recursive Modeling Method proposed before in [4]. RMM endows an agent with a compact specialized representation of other agents beliefs, abilities, and intentions. 1 As such, it allows the agent to predict the message s decision theoretic (DT) pragmatics, i.e. how a particular message will change the decision making situation of the agents, ....

....like the one in Figure 4, can be solved using dynamic programming. The solution proceeds bottom up and results in expected utility values being assigned to Battery2 s alternative behaviors, the best of which can then be executed. For a more detailed solution procedure and further discussion, see [4, 10]. 3 Communication in Anti Air Defense We identify a communicative act with its decision theoretic (DT) pragmatics, defined as the transformation of the state of knowledge about the decision making situation (i.e. the recursive model structure) the act brings about. We model DT pragmatics using ....

P. J. Gmytrasiewicz and E. H. Durfee. A rigorous, operational formalization of recursive modeling. In Proceedings of the First International Conference on Multi-Agent Systems, pages 125--132, Menlo Park, 1995. AAAI Press/The MIT Press.

....the hostile missiles. Based on these attributes combined, each unit has to determine the optimal action from his probabilistic decision model. For the purpose of coordinated decision making in a multiagent environment, our research uses the Recursive Modeling Method (RMM) previously reported in Gmytrasiewicz and Durfee (1995) and Gmytrasiewicz (1996) RMM enables an agent to model the other agents and to rationally coordinate with them even if no protocol or overall plan can be established explicitly in advance. Using RMM as a decision making tool, an agent rationally selects his action under uncertainty guided by the ....

....Agent To be rational in decision theoretic sense, the agents follow the principle of maximum expected utility (PMEU) Russell and Norvig, 1995, chap. 14) In this section, we will show how PMEU can be implemented in the antiair defense domain using the Recursive Modeling Method (RMM) RMM (Gmytrasiewicz and Durfee, 1995, and Gmytrasiewicz, 1996) will be used to model the other agent, and to select the most appropriate missile to intercept by a given defense battery. 2 3.1 An Example Scenario Our approach is to take the agent oriented perspective. In the example scenario (Figure 1) we view the decision making ....

Gmytrasiewicz, P. J., and Durfee, E. H. (1995). A rigorous, operational formalization of recursive modeling.

....and predict consumer and customer behavior. However, it has begun to be widely used in AI applications only recently [3] even though formal notions of rationality defined in the field of decision theory. Our approach to modeling uses the Recursive Modeling Method (RMM) previously reported in [5]. RMM enables an agent to model the other agents and to rationally coordinate with them even if no protocol or overall plan can be established explicitly in advance. Using RMM as a decision making tool, an agent rationally selects his action under uncertainty guided by the principle of expected ....

.... While in Shoham s work the behavior of an agent is guided by pre defined rules, our agents rationally determine how to behave based on the principle of maximum expected utility (PMEU) 14] PMEU can be implemented in the anti air defense domain using the Recursive Modeling Method (RMM) RMM [5] will be used to model other agents and to coordinate with them. Further, Bayesian belief update will be applied to recognize the other agents capabilities correctly. belief (frame) decision (RMM) Agent Bayesian belief update Information Act Environment and Other Agents capability (frame) ....

[Article contains additional citation context not shown here]

P. J. Gmytrasiewicz and E. H. Durfee. A rigorous, operational formalization of recursive modeling. In Proceedings of the First International Conference on Multi-Agent Systems, pages 125--132, Menlo Park, 1995. AAAI Press/The MIT Press.

No context found.

P.J. Gmytrasiewicz and E.H. Durfe. A rigorous, operational formalization of recursive modeling. In Proceedings of the First International Conference on Multi-Agent Systems (ICMAS), pp. 125--132, 1995.

No context found.

P.J. Gmytrasiewicz and E.H. Durfe. A rigorous, operational formalization of recursive modeling. In Proceedings of the First International Conference on Multi-Agent Systems (ICMAS), pp. 125--132, 1995.

No context found.

P. J. Gmytrasiewicz, and E. H. Durfee.: A rigorous, operational formalization of recursive modeling. In Proc. First International Conference on Multiagent Systems. San Francisco, California (1995) 125--132.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC