P. J. Gmytrasiewicz, E. H. Durfee, and D. K. Wehe. A decision-theoretic approach to coordinating multi-agent interations. In Proc. Int. Joint Conf. on Artif. Intell., pages 62-68, 1991.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....Savage [77] who simultaneously axiomatized utility and subjective behavior) This is also in keeping with a large (and growing) body of work within artificial intelligence that attributes rationality (or explores the consequences of attributing rationality) to autonomous agents. See, for example, [46, 25, 70, 12, 42, 34]. 1.3 Overview of this Article In this article we present a method for reaching consensus based on the Clarke Tax mechanism [6, 7] CTm) and consider how this mechanism could be used among rational automated agents. Parts of this work have appeared previously in [18, 22, 20, 19] In Section ....

....that maximizes the social utility (minimizes the damage according to the group s perspective) In addition, there is no need to assume that the agents are benevolent. 38 One game theoretic method for coordinating the activities of autonomous agents, within MAS, is the Recursive Modeling Method [34]. Each agent models the other agents in a recursive manner and thus acquires probabilistic knowledge about the expected utility values that the other agents have about their preferences, abilities, and the world. Each agent looks for an action that will maximize its individual utility, by ....

Piotr J. Gmytrasiewicz, Edmund H. Durfee, and David K. Wehe. A decision-theoretic approach to coordinating multiagent interactions. In Proceedings of the Twelfth International Joint Conference on Artificial Intelligence, pages 62--68, August 1991.

....and his colleagues [14] have developed a logic of knowledge and belief to model multiagent coordination. Their framework permits an agent to reason not only about the world and its own actions, but also to simulate and model the behavior of other agents in the environment. In a separate paper [15], they show how one agent can reason with a probabilistic view of the behavior of other agents so as to achieve coordination. 7. ....

P. Gmytrasiewicz, E. Durfee, and D. Wehe. A Decision-Theoretic Approach to Coordinating Multiagent, Interactions. In Proceedings of the 12th IJCAI, pages 62-68, Sydney, Australia, 1991. Morgan Kaufmann.

....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 ....

Gmytrasiewicz P., Durfee E., Wehe D. A DecisionTheoretic Approach to Coordinating Multiagent Interactions. In Proceedings of IJCAI'91, pp. 62-68, 1991.

....it possible to move the ball quickly to another part of the eld and outplay the opponent s defense. 6 Related Work Two other approaches that model the behavior of other agents future actions are RMM (recursive modeling method) and IMBBOP ( ideal model based behavior outcome prediction ) RMM [6] provides a theoretical framework for representing and using the knowledge that an agent has about its expected payo s and those of others. An agent models the internal state and action selection strategy of the other agents in order to predict its action. The method is recursive since the other ....

P. Gmytrasiewicz, E. Durfee, and D. Wehe. A decision-theoretic approach to coordinating multiagent interactions. In Proceedings of the Twelfth International Joint Conference on Arti cial Intelligence, pages 62-68, 1991.

....technique, which we call ideal model based behavior outcome prediction (IMBBOP) Our technique also includes a method for relaxing this optimality assumption. An alternative to minimax, in which other agents aren t necessarily assumed to act optimally, is the recursive modeling method (RMM) [3]. Using RMM, an agent models the internal state and action selection strategy of another agent in order to predict its actions. This method is recursive because the other agent might similarly be modeling the original agent, leading to an arbitrary depth of reasoning (techniques for limiting this ....

P. J. Gmytrasiewicz, E. H. Durfee, and D. K. Wehe. A decision-theoretic approach to coordinating multiagent interactions. In Proceedings of the Twelfth International Joint Conference on Artificial Intelligence, pages 62--68, 1991.

....in relation to its theoretical optimal actions in our model based technique, IMBBOP. Our technique also includes a method for relaxing this optimality assumption. An alternative to minimax, in which other agents aren t necessarily assumed to act optimally, is the recursive modeling method (RMM) (Gmytrasiewicz, Durfee, Wehe 1991). Using RMM, an agent models the internal state and action selection strategy of another agent in order to predict its actions. This method is recursive because the other agent might similarly be modeling the original agent, leading to an arbitrary depth of reasoning (techniques for limiting this ....

Gmytrasiewicz, P. J.; Durfee, E. H.; and Wehe, D. K. 1991. A decision-theoretic approach to coordinating multiagent interactions. In Proceedings of the Twelfth International Joint Conference on Artificial Intelligence, 62--68.

....decision maker knows all game components, including the loss functionals, of other decision makers. Another item could be introduced that reflects imperfect information that each decision maker has about the game itself. Problems of this type are quite realistic, yet are very difficult to model [73], 83] The information of each decision maker could be represented, for example, as a probability density over a set of possible games. To make appropriate decisions, each decision maker must speculate about the knowledge that other decision makers have regarding the game. This type of ....

....This type of second guessing can progress for an infinite number of layers, which leads to a formidable modeling task. One approach to problems of this type is the Recursive Modeling Method, which finds strategies that are optimal in the expected sense by averaging over a finite number of layers [73]. 5.3 Unifying the Concepts from Chapters 2 4 This section specializes the structure from Section 5.2 to minimally encompass the essential concepts from Chapters 2 through 4. A multiple robot motion planning problem is defined in which each robot experiences uncertainty in configuration ....

[Article contains additional citation context not shown here]

P. J. Gmytrasiewicz, E. H. Durfee, and D. K. Wehe. A decision-theoretic approach to coordinating multi-agent interations. In Proc. Int. Joint Conf. on Artif. Intell., pages 62--68, 1991.

....in relation to its theoretical optimal actions in our model based technique, IMBBOP. Our technique also includes a method for relaxing this optimality assumption. An alternative to minimax, in which other agents aren t necessarily assumed to act optimally, is the recursive modeling method (RMM) (Gmytrasiewicz, Durfee, Wehe 1991). Using RMM, an agent models the internal state and action selection strategy of another agent in order to predict its actions. This method is recursive because the other agent might similarly be modeling the original agent, leading to an arbitrary depth of reasoning (techniques for limiting this ....

Gmytrasiewicz, P. J.; Durfee, E. H.; and Wehe, D. K. 1991. A decision-theoretic approach to coordinating multiagent interactions. In Proceedings of the Twelfth International Joint Conference on Articial Intelligence, 62-68.

....an agent s self model and its recursive self model in service of deception and other actions. One key issue for future work is understanding the broader applicability of these lessons. To this end, we plan to explore the relationships of our approach with formal methods for recursive agent modeling(Gmytrasiewicz, Durfee, Wehe 1991; Wilks Ballim 1987) This may help generalize the tracking approach introduced in this paper to other multi agent environments, including ones for entertainment or education. ....

Gmytrasiewicz, P. J.; Durfee, E. H.; and Wehe, D. K. 1991. A decision theoretic approach to co-ordinating multi-agent interactions. In Proceedings of International Joint Conference on Artificial Intelligence.

.... [26] but is influenced by them, and by the importance of environmental uncertainty and dependency that appear in contingency theoretic and open systems views of organizations [22, 14, 40, 34] As a problem representation for computational tasks, it is richer and more expressive than game theory [32, 45, 18] or team theory [20] representations. For example, a typical game or team theory problem statement is concerned with a single decision; a typical T MS objective problem solving episode represents the possible outcomes of many sequences of choices that are interrelated with one another (e.g. ....

Piotr J. Gmytrasiewicz, Edmund H. Durfee, and David K. Wehe. A decision-theoretic approach to coordinating multiagent interactions. In Proceedings of the Twelfth International Joint Conference on Artificial Intelligence, pages 62--68, Sydney, Australia, August 1991.

....(f 2 ; f 1 ; f 0 ; NIL) is one that uses a strategy f 2 , and has a model of its opponent, f 1 ; f 0 ; NIL) The opponent s model uses a strategy f 1 and has a model, f 0 ; NIL) of the player. The recursive definition of a player is in the spirit of the Recursive Modeling Method by Gmytrasiewicz, Durfee and Wehe (1991). M receives a position, a depth limit, and a player, and outputs a move selected by the player and its value. The algorithm generates the successor boards and simulates the opponent s search from each of them in order to anticipate its choice. This simulation is achieved by applying the ....

Gmytrasiewicz, P. J.; Durfee, E. H.; and Wehe, D. K. 1991. A decision theoretic approach to coordinating multiagent interactions. In Proceedings of the Twelfth International Joint Conference on Artificial Intelligence (IJCAI 91), 62 -- 68.

....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, E. H. Durfee, and D. K. Wehe. A decision-theoretic approach to coordinating multiagent interactions. In Proceedings of the Eleventh International Joint Conference on Artificial Intelligence, pages 166--172, July 1991.

....and Durfee (1992) have developed a logic of knowledge and belief to model multiagent coordination. Their framework permits an agent to reason not only about the world and its own actions, but also to simulate and model the behavior of other agents in the environment. In a separate paper (P. Gmytrasiewicz and Wehe. 1991), they show how one agent can reason with a probabilistic view of the behavior of other agents so as to achieve coordination. This is good work. There are some significant differences between our work and theirs. First, we focus on agents that are built on top of arbitrary data structures. Second, ....

P. Gmytrasiewicz, E. D. and D. Wehe. (1991). A Decision-Theoretic Approach to Coordinating Multiagent Interactions. In Proceedings of IJCAI 1991, pp. 62--68. Morgan Kaufman.

....one that uses a strategy S 2 , and has a model of its opponent, S 1 ; S 0 ; NIL) The opponent s model uses a strategy S 1 and has a model, S 0 ; NIL) of the player. The recursive definition of a player is in the spirit of the Recursive Modelling Method (RMM) by Gmytrasiewicz, Durfee and Wehe [15]. 2.1 The M algorithm Most of the game playing programs use a minimax search procedure in which the player evaluates boards by a function f , and believes that the opponent evaluates boards by the function Gammaf . Assume that the player uses a function f 1 , but believes that the opponent ....

E. H. D. P. J. Gmytrasiewicz and D. K. Wehe. A decision theoretic approach to coordinating multiagent interactions. In Proceedings of the International Joint Conference on Artifical Intelligence (IJCAI 91), pages 62--68, 1991.

....military agreements. It may be possible to adapt her structure and some of her negotiation heuristics into an automated bargaining agent. We might also adopt Gmytrasiewicz et al. s recursive modeling method so our agent could use its predictions of other agents responses when planning its actions [5,5]. In addition, Kraus et al. presented a strategic model of negotiation that takes time into account during the negotiation process [9] Such mechanism could be applicable to our experiments as they get more dynamic and time sensitive. 7 Conclusion In this paper, we advocated a new challenging ....

P. J. Gmytrasiewicz, E. H. Durfee, and D. K. Wehe. A decision-theoretic approach to coordinating multiagent interactions. In Proc. 12th Int. Joint Conf. on A.I., pages 62--68, August 1991.

....f 0 , for determining the opponent s choices. Note that the opponent is a maximizer. Part (b) shows the recursive calls of M 1 using f 1 , for evaluating the opponent choices. The recursive definition of a player is in the spirit of the Recursive Modeling Method by Gmytrasiewicz, Durfee and Wehe [6]. The M algorithm returns the M n value of a game state. It receives a position, a depth limit, and an n level player 2 , and determines the M n value of the game position and the move selected by the player. The algorithm simulates the opponent s search for each successor in order to ....

P. J. Gmytrasiewicz, E. H. Durfee, and D. K. Wehe. A decision theoretic approach to coordinating multiagent interactions. In Proceedings of the Twelfth International Joint Conference on Artificial Intelligence (IJCAI 91), pages 62 -- 68, Sydney Australia, August 1991.

.... [Burton and Obel, 1984, Galbraith, 1977] some computer scientists, both in DAI and in distributed processing, use market coordination mechanisms [Malone et al. 1983, Wellman, 1993] the more complex game theoretic mechanisms [Rosenschein and Genesereth, 1985, Zlotkin and Rosenschein, 1991, Gmytrasiewicz et al. 1991] or team theory [Ho, 1980] The older schools of thought invariably use either quantitative or qualitative methods of description for the task environment in which the agents under study are immersed. Sociologists need descriptions of the environment to explain Why . Neoclassical economists ....

....1990] but it would be very hard to state good featural characterizations using a simulator. The second input to the form of the model is the mathematically formal work in DAI (for example, Genesereth et al. 1986, Rosenschein and Genesereth, 1985, Malone, 14 1987, Cohen and Levesque, 1990, Gmytrasiewicz et al. 1991] There was no reason that the complexities that occur in our earlier DVMT simulation work could not be given clear semantics. The existing formalisms, however, universally eschew much complexity to allow for optimal analyses to be carried through. For example, Malone [Malone, 1987] formalizes ....

[Article contains additional citation context not shown here]

Piotr J. Gmytrasiewicz, Edmund H. Durfee, and David K. Wehe. A decision-theoretic approach to coordinating multiagent interactions. In Proceedings of the Twelfth International Joint Conference on Artificial Intelligence, pages 62--68, Sydney, Australia, August 1991.

....AP may deceive the opponent into believing that it has fired a missile without firing one. Alternatively, as the opponent attempts to deceive AP, in some situations, AP may discover this deception. This topic is 24 Computational Intelligence related to issues of recursive agent modeling(Gmytrasiewicz, Durfee, Wehe 1991; Wilks Ballim 1987) Work on this topic is already underway. A fifth issue for future work is extending event tracking to situations involving different groups of agents. Key questions that come up here include: how does a WCPS generalize to this situation Should an agent use a single WCPS or ....

Gmytrasiewicz, P. J., E. H. Durfee and D. K. Wehe 1991. A decision theoretic approach to co-ordinating multi-agent interactions. In Proceedings of International Joint Conference on Event Tracking in a Dynamic Multi-agent Environment 25 Artificial Intelligence.

....intentions depends in part on its commitment to the beliefs that support that intention. However, commitment also depends on the importance of the goal state, and the likelihood and difficulty of achieving the goal state. Here we borrow from decision theoretic approaches to action choice (e.g. [9]) though we will make no assumptions about the availability of numerical estimators of utility. Instead, we extend our notion of endorsements on beliefs to apply to intentions by including heuristic descriptions of the utility of goal states and the effort required to perform the actions leading ....

P. J. Gmytrasiewicz, E. H. Durfee and D. K. Wehe, A Decision Theoretic Approach to Coordinating Multi-Agent Interactions. In: Proceedings of IJCAI-91, Sydney, 1991.

....maker knows all game components, including the loss functionals, of other decision makers. Another sensing model could be introduced that reflects imperfect information that each decision maker has about the game itself. Problems of this type are quite realistic, yet are very difficult to model [45, 49]. The information of each decision maker could be represented, for example, as a pdf over a set of possible games. To make appropriate decisions, each decision maker must speculate about the knowledge that other decision makers have regarding the game. This type of second guessing can progress for ....

P. J. Gmytrasiewicz, E. H. Durfee, and D. K. Wehe. A decision-theoretic approach to coordinating multi-agent interations. In Proc. Int. Joint Conf. on Artif. Intell., pages 62--68, 1991.

....transforms team plans into separate role plans for execution by individuals, with rigidly embedded communications. STEAM purposely avoids such transformations, so agents can flexibly reason with (i) explicit team goals plans; and (ii) selective communication (seen to be important in practice) In (Gmytrasiewicz, Durfee, Wehe, 1991), decision theory is applied for message prioritization in coordination based on the agents recursive modeling of each others actions. STEAM applies decision theory for communication selectivity and enhancements, but in a very different context practical operationalization of general, ....

Gmytrasiewicz, P. J., Durfee, E. H., & Wehe, D. K. (1991). A decision theoretic approach to co-ordinating multi-agent interactions. In Proceedings of International Joint Conference on Artificial Intelligence.

....to represent and reason about actions, plans and knowledge of other agents in order to coordinate with them. The ability to model another agent s goals and beliefs has a direct impact on an agent s ability to reason about other agents and consequently make it a better team player. Methods like RMM[Gmytrasiewicz et al. 1991] have been proposed where each agent models the other agents at multiple levels of recursive reasoning involving, for example, knowledge of the other agent about the present agent and so on. These methods highlight the need for an agent to model other agents for better coordination. However, the ....

Gmytrasiewicz, P. J.; Durfee, E. H.; and Wehe, D. K. A decisiontheoretic approach to coordinating multiagent interactions. In Proceedings of the Twelfth International Joint Conference on Artificial Intelligence, pages 62--68, Sydney, Australia, August 1991.

.... is applicable to many areas dealing with complex structures e.g. ffl computer aided manufacturing or computer aided design [1,4,11,30,38] where a complex object=a final artifact (assembly) is produced (designed) from inventory (elementary) parts by a dedicated team of agents; ffl logistics [10,11,14,24] where complex structures are organized from existing elementary structures (units) to perform a task according to a given specification; ffl adaptive control of complex systems [12,17,27] where the task consists in maintaining a given constraint (specification) by adaptive adjustment of ....

....etc. ffl business re engineering [2,24] where the task is to adaptively modify a complex object (structure, organization, resources, etc. according to the current economic situation (specification) ffl cooperative distributed problem solving including planning, dynamic task assignment etc. [5,6,7,8,9,10,11,13,15,30,37,38,40,41,46,47] where the task is to organize a system of agents into a scheme of local teams for solving a problem (specification) ffl automated fabrication [4] where the task is to build complex objects (e.g. mechanisms) by layer after layer synthesis; ffl preliminary stage of design process [30] where the ....

[Article contains additional citation context not shown here]

Gmytrasiewicz P.J., Durfee E.H., Wehe D.K.: A decision- theoretic approach to coordinating multi-agent interactions, IJCAI-91, San Mateo:Morgan Kaufmann 1991, 62-68

....presents the results of an implementation of our algorithm, from which we derive some conclusions and directions for further work, discussed in the Conclusion. RMM and the Pursuit Task: The basic mod1 eling primitives we use are based on the Recursive Modeling Method (RMM) Gmytrasiewicz 1992; Gmytrasiewicz, Durfee, Wehe 1991; Durfee, Gmytrasiewicz, Rosenschein 1994; Durfee, Lee, Gmytrasiewicz 1993) RMM provides a theoretical framework for representing and using the knowledge that an agent has about its expected payo#s and those of others. To use RMM, an agent is expected to have a payo# matrix where each entry ....

Gmytrasiewicz, P. J.; Durfee, E. H.; and Wehe, D. K. 1991. A decision-theoretic approach to coordinating multiagent interactions. In Proceedings of the Twelfth International Joint Conference on Artificial Intelligence.

....identical to the ones we are dealing with, but the emphasis is on the communication channel design, as opposed to the decision making of the individual agents motivated by maximization of the expected utility of the messages exchanged. Our approach is based on the Recursive Modeling Method (RMM) [4], and on our analysis of how communication transforms the RMM hierarchy [5] outlined also in the next section) which is intended to be a complete representation of an agent s knowledge relevant to the decision making process in a multiagent environment. As in our earlier work, the main guideline ....

.... We get here a recursive pattern on the communicative level, and the resulting recursive hierarchy, which we call a communication hierarchy, can be solved by methods similar to those used in the case of recursive hierarchies containing physical actions, called action hierarchies and analyzed in [4, 5]. However, as it turns out, each of the recursive levels of communicative options of the agents requires solution of at least one action hierarchy, as opposed to the solution of a single payoff matrix in action hierarchies. In addition, solving a communication hierarchy introduces the notion of ....

[Article contains additional citation context not shown here]

Piotr J. Gmytrasiewicz, Edmund H. Durfee, and David K. Wehe. A decision-theoretic approach to coordinating multiagent interactions. In Proceedings of the Twelfth International Joint Conference on Artificial Intelligence, pages 62--68, August 1991.

....the air combat simulation environment, while Carmel and 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 ....

Gmytrasiewicz, P. J., Durfee, E. H., and Wehe, D. K. (1991). A decision-theoretic approach to coordinating multiagent interactions. In Proceedings of the Eleventh International Joint Conference on Artificial Intelligence, 166--172.

No context found.

Gmytrasiewicz, P. J.; Durfee, E. H.; and Wehe, D. K. 1991a. A decision-theoretic approach to coordinating multiagent interactions. In Proceedings of the Twelfth International Joint Conference on Artificial Intelligence, 62--68.

....model they think the defenders have of them on the third level (see Figure 1a) and so on. The resulting hierarchy has one branch of alternating models representing the decision making situations of the sides via their payoff matrices. As we have argued in other applications of the RMM algorithm [9, 8], the hierarchy in Figure 1a is finite and has to end, although at a possibly deep and unknown level. That is due to the fact that the two sides do not have any practical means for arriving at knowledge that would be nested down to infinity. A very elegant proof of this has been given by Halpern ....

....the character of our prototypical interactions in a way not considered before. The threatening agent must wait for the action of the threatened agent, and only if this action is the action B, can the threat be enforced. Thus, unlike in the previous case above (and the other cases considered in [8, 9]) the actions of the players cannot be considered as simultaneous. 4.1 Verbal Threat We will again use the scenario of the defending and the invading agents presented above to illustrate our approach. However, to make our presentation less complex, we will assume both sides have only two ....

[Article contains additional citation context not shown here]

Piotr J. Gmytrasiewicz, Edmund H. Durfee, and David K. Wehe. A decision-theoretic approach to coordinating multiagent interactions. In Proceedings of the Twelfth International Joint Conference on Artificial Intelligence, pages 62--68, August 1991.

....number of levels of recursive modeling. In the Recursive Modeling Method (RMM) for example, we have developed a procedure whereby agents can build this recursive nesting of models and can use the resultant modeling hierarchy to infer rational courses of action for others and for themselves [ Gmytrasiewicz et al. 1991 ] RMM provides a rigorous foundation for such decisionmaking situations, based on RMM s concept of a solution. However, as we show in this paper, RMM s original formulation leads to decisions that differ from those that would be derived by traditional game theoretic methods, because those ....

....mixed strategies, and if so, how is that selection done. Mixed Strategies Through RMM Key functions in a much simplified version of RMM, which does not consider possible horizontal branching representing uncertainty about alternative payoff matrices other agents might subscribe to (see [ Gmytrasiewicz et al. 1991 ] are shown in Figure 3. Note that this example implementation uses a very simple method to change the value of k at successively lower levels of the hierarchy: it multiplies the value of k at the previous level by a constant (less than 1) This approach allows the algorithm to asymptotically ....

Gmytrasiewicz, Piotr J.; Durfee, Edmund H.; and Wehe, David K. 1991. A decision-theoretic approach to coordinating multiagent interactions. In Proceedings of the Twelfth International Joint Conference on Artificial Intelligence.

.... Perrault 1979; Durfee Montgomery 1990) Communication poor coordination techniques do exist, including social conventions (Shoham Tennenholtz 1992) focal points (Kraus Rosenschein 1991) decision theoretic (Genesereth, Ginsberg, Rosenschein 1984) and game theoretic recursive modeling (Gmytrasiewicz, Durfee, Wehe 1991). In general, these techniques emphasize implicitly or explicitly infering others actions based on established norms for behavior or on beliefs about the preferences or interests of others. Thus, social conventions constrain behavior to make others predictable, but can be overconstraining in ....

Gmytrasiewicz, P. J.; Durfee, E. H.; and Wehe, D. K. 1991. A decision-theoretic approach to coordinating multiagent interactions. In Proceedings of the Twelfth International Joint Conference on Artificial Intelligence.

....the rational one in the situation at hand [8] Within our approach in RMM this parameter is formally introduced into the solution method in the section below. 3 The Formalizm of Recursive Modeling In this section we provide a brief overwiev of the Recursive Modeling Method previously presented in [18, 19, 20]. The Recursive Modeling Method consists of a modeling structure that represents the an agent s knowledge in all of its nested levels, and the solution method that traverses the structure equivalent to the original modeling structure to arrive at the rational choice of an agent s action in a ....

....In both cases, if the intentionality and rationality was not used to arrive at these predictions, they can be treated as sub intentional models within the RMM framework. The definition of the recursive model structure and the intentional model are recursive, but, as we argued in more detail in [18, 19, 20], and as suggested by Halpern and Moses in [22] the recursion is bound to end due to practical limitations in attaining infinite knowledge. Intuitively, agents that have interacted for a finite amount of time could possibly have exchanged only a finite number, say , of messages. Therefore, the ....

Piotr J. Gmytrasiewicz, Edmund H. Durfee, and David K. Wehe. A decision-theoretic approach to coordinating multiagent interactions. In Proceedings of the Twelfth International Joint Conference on Artificial Intelligence, pages 62--68, August 1991.

.... if the agents see each other as rational, then they can make a deal allowing them to cooperate [Rosenschein and Genesereth, 1985] Alternatively, they can apply meta game techniques to evaluate the performance of alternative strategies to select the cooperative strategy from a purely selfish view [Gmytrasiewicz et al. 1991], leading to self organization for their mutual benefit. Organization theory [Malone, 1987] has also looked at the problem of how organizations form and why individuals are willing to become parts of an organization. In joining an organization, an individual foregoes some of his freedom because ....

Gmytrasiewicz, Piotr J.; Durfee, Edmund H.; and Wehe, David K. 1991. A decision-theoretic approach to coordinating multiagent interactions. In Proceedings of the Twelfth International Joint Conference on Artificial Intelligence.

No context found.

Gmytrasiewicz, P. J.; Durfee, E. H.; and Wehe, D. K. 1991b. A decision-theoretic approach to coordinating multiagent interactions. In Proceedings of the Twelfth International Joint Conference on Artificial Intelligence, 62--68.

....In this paper we present a set of interrelations among elements of the philosophical, theoretical and implementational levels that, we hope, could be seen as a contribution toward such a unifying view. Our considerations draw on our recent work in Distributed Artificial Intelligence [8, 9], which has been directed toward creating a rigorous foundation for addressing problems of 0 This research was supported, in part, by the Golda Meir Fellowship and the Alfassa Fund administered by the Hebrew University of Jerusalem, and by the National Science Foundation under grant IRI 9015423 ....

....agents behavior contained in the intentional stance and other elements of Dennett s ladder of personhood, as formalized in game theory. G1, worth 2 G2, worth 4 R1 R2 Cost = 1 Cost = 2 Cost = 2 Cost = 1 Wall Figure 3: Another Example of Multiagent Interaction. RMM is described in more detail in [8, 9]; now we just review the simple example of interaction in Figure 3 and concentrate on RMM s relations with previously developed notions. The scenario in Figure 3 is similar to the one in Figure 1, but we have added the element of uncertainty. Now agent R1 notices that there is an obstacle between ....

[Article contains additional citation context not shown here]

Piotr J. Gmytrasiewicz, Edmund H. Durfee, and David K. Wehe. A decision-theoretic approach to coordinating multiagent interactions. In Proceedings of the Twelfth International Joint Conference on Artificial Intelligence, pages 62--68, August 1991.

....the alternative branches labeled with the probabilities R 1 assigns to each of the models being correct. The third level is occupied by uniform probability distributions representing R 1 s lack of knowledge of how it is being modeled by R 2 . Solving the Example As has been summarized elsewhere [9], the recursive model generated by RMM can be solved so as to determine a rational action for the agent represented at the root of the tree, based on expected rational actions taken on the part of agents modeled deeper in the tree. Thus, the solution method proceeds from the leaves upward. In the ....

....of threats is that they change the character of our prototypical interactions in a way not considered before. The threatening agent must wait for the action of the threatened agent, and take its own action accordingly. Thus, unlike in the previous case above (and the other cases considered in [9, 10]) the actions of the players cannot be considered as simultaneous. Threats will be assumed to have the form If you do A, then I will do B, where A is an option of an opponent and B is an option available to the threatening agent. One of the subtleties of threats, as discussed for example in ....

Piotr J. Gmytrasiewicz, Edmund H. Durfee, and David K. Wehe. A decision-theoretic approach to coordinating multiagent interactions. In Proceedings of the Twelfth International Joint Conference on Artificial Intelligence, August 1991.

....static structuring mechanisms. At the other extreme are researchers who strive toward endowing each agent in a complex system with the ability to dynamically maintain models of other agents and to use these models to make arbitrarily complex coordination decisions interleaved with execution (e.g. [Gmytrasiewicz et al. 1991]) Between these extremes are the many techniques for coordinating agents within the bounds of some fixed commitments upon which they can all depend (e.g. Decker, 1995; Durfee et al. 1987; Gasser et al. 1989] But these techniques have generally been developed to meet the needs of particular ....

Gmytrasiewicz, P., Durfee, E., and Wehe, D., A decision-theoretic approach to coordinating multiagent interactions." In IJCAI-91, pp 62-68.

....presents the results of an implementation of our algorithm, from which we derive some conclusions and directions for further work, discussed in the Conclusion. RMM and the Pursuit Task: The basic mod eling primitives we use are based on the Recursive Modeling Method (RMM) Gmytrasiewicz 1992; Gmytrasiewicz, Durfee, Wehe 1991; Durfee, Gmytrasiewicz, Rosenschein 1994; Durfee, Lee, Gmytrasiewicz 1993) RMM provides a theoretical framework for representing and using the knowledge that an agent has about its expected payoffs and those of others. To use RMM, an agent is expected to have a payoff matrix where each entry ....

Gmytrasiewicz, P. J.; Durfee, E. H.; and Wehe, D. K. 1991. A decision-theoretic approach to coordinating multiagent interactions. In Proceedings of the twelfth international joint conference on artificial intelligence.

....in their paper Cooperation and Conflict Resolution via Negotiation Among Autonomous Agents in Non Cooperative Domains, also employ formal methods for investigating coordination among agents. Their approach takes a decision theoretic, probablistic view of decisionmaking in multiagent environments [24, 36, 50], where each agent is acting so as to maximize its expected utility. By considering the worths of their goals and the costs of achieving them, the interaction between agents can be characterized as requiring cooperation, compromise, or conflict. Moreover, by utilizing probablistic methods, the ....

Piotr J. Gmytrasiewicz, Edmund H. Durfee, and David K. Wehe. A decision-theoretic approach to coordinating multiagent interactions. In Proceedings of the Twelfth International Joint Conference on Artificial Intelligence, August 1991.

....identical to the ones we are dealing with, but the emphasis is on the communication channel design, as opposed to the decision making of the individual agents motivated by maximization of the expected utility of the messages exchanged. Our approach is based on the Recursive Modeling Method (RMM) [4], and on our analysis of how communication transforms the RMM hierarchy [5] outlined also in the next section) which is intended to be a complete representation of an agent s knowledge relevant to the decision making process in a multiagent environment. As in our earlier work, the main guideline ....

.... We get here a recursive pattern on the communicative level, and the resulting recursive hierarchy, which we call a communication hierarchy, can be solved by methods similar to those used in the case of recursive hierarchies containing physical actions, called action hierarchies and analyzed in [4, 5]. However, as it turns out, each of the recursive levels of communicative options of the agents requires solution of at least one action hierarchy, as opposed to the solution of a single payoff matrix in action hierarchies. In addition, solving a communication hierarchy introduces the notion of ....

[Article contains additional citation context not shown here]

Piotr J. Gmytrasiewicz, Edmund H. Durfee, and David K. Wehe. A decision-theoretic approach to coordinating multiagent interactions. In Proceedings of the Twelfth International Joint Conference on Artificial Intelligence, pages 62--68, August 1991.

....strong motivation for agents being nice is that agents might encounter each other repeatedly, and so they might be punished for past transgressions. An agent might thus determine that its long term payoff will be better if it does not antagonize another. This intuition has been captured [ Gmytrasiewicz et al. 1991a; Vane and Lehner, 1990 ] and has introduced the game theoretic definition of cooperation as what agents will do if they expect to interact infinitely many times into DAI. Put Things Back Where You Found Them Agents that share a world must contend with the dynamics that each introduces to the others. Given ....

Gmytrasiewicz, Piotr J.; Durfee, Edmund H.; and Wehe, David K. 1991a. A decision-theoretic approach to coordinating multiagent interactions. In Proceedings of the Twelfth International Joint Conference on Artificial Intelligence.

No context found.

P. J. Gmytrasiewicz, E. H. Durfee, and D. K. Wehe. A decision-theoretic approach to coordinating multi-agent interations. In Proc. Int. Joint Conf. on Artif. Intell., pages 62-68, 1991.

No context found.

P. J. Gmytrasiewicz, E. H. Durfee, and D. K. Wehe. A decision-theoretic approach to coordinating multi-agent interations. In Proc. Int. Joint Conf. on Artif. Intell., pages 62--68, 1991.

No context found.

P. J. Gmytrasiewicz, E. H. Durfee, and D. K. Wehe. A decision-theoretic approach to coordinating multi-agent interations. In Proc. Int. Joint Conf. on Artif. Intell., pages 62--68, 1991.

No context found.

P. Gmytrasiewicz, E. Durfee, and D. Wehe. A decision-theoretic approach to coordinating multiagent interactions. In Proceedings of the Twelfth International Joint Conference on Artificial Intelligence, pages 62--68, 1991.

No context found.

P. Gmytrasiewicz, E. Durfee, and D. Wehe. A Decision-Theoretic Approach to Coordinating Multiagent Interactions. In Proceedings of IJCAI 1991, pages 62-68. Morgan Kaufman, 1991.

No context found.

P. Gmytrasiewicz, E. Durfee, and D. Wehe. A Decision-Theoretic Approach to Coordinating Multiagent Interactions. In Proceedings of IJCAI 1991, pages 62--68. Morgan Kaufman, 1991.

No context found.

P. Gmytrasiewicz, E. Durfee, and D. Wehe. A Decision-Theoretic Approach to Coordinating Multiagent Interactions. In Proceedings of IJCAI 1991, pages 62--68. Morgan Kaufman, 1991.

No context found.

Gmytrasiewicz, P. J., Durfee, E. H., & Wehe, D. K. (1991a). A decision theoretic approach to coordinating multiagent interaction. In Proceedings of the Twelfth International Joint Conference on Artificial Intelligence, pp. 62--68 Sydney, Australia.

No context found.

Gmytrasiewicz91, P.J., Durfee, E.H., & Wehe, D.K. (1991). A decision-theoretic approach to coordinating multiagent interactions. In Proceedings of the Twelfth International Joint Conference on Artificial Intelligence, pages 62--68.

No context found.

Gmytrasiewicz, P. J., Durfee, E. H. and Wehe, D. K. A Decision- Theoretic Approach to Coordinating Multiagent Interactions. In Proceedings of the Twelfth International Joint Conference On Artificial Intelligence (Sydney, Australia 1991). International Joint Conferences on Artificial Intelligence, Inc. pp. 62-68.

First 50 documents  Next 50

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