Fifth International Conference on Fuzzy Systems and Knowledge Discovery
An Outcome Preference Information Aggregation Model
and Its Algorithm in Hypergame Situations
Qu Yong, Song Yexin, Zhang Jianjun
College of Science, Naval University of Engineering,
Wuhan, Hubei 430033, China
qqqyyy0821@21cn.com,yxsong@21cn.com,wahh0912@163.com
In a simple game, there are no misperceptions, each player is represented by only one PV, and all the players
see the same set of PVs. However, in real situations, it is In hypergame situations, it is important for a player to
get the more correct opponent players’ outcome quite usual that the players involved perceive the conflict
differently and subjectively. Hypergame [2,3,4] is a preference information. In this paper, based on the
typical framework to deal with such situations. Contrary principle of fuzzy pattern recognition, a nonlinear
programming model is established for integrating to the traditional game theory, the players involved in a opponent players’ different outcome preference hypergame may have incorrect subjective perception
about the other players’ options, strategies, or preferences, evaluation values perceived by different experts without
or even be unaware of some of the players in the game. In weight information. An iteration algorithm for solving the
this paper, we only consider such hypergame models model is developed. Using the proposed model and its
where each player correctly perceives the players in the algorithm, not only the weight of each expert but also the
game and the strategy set of the opponent players, but the integrated outcome preferences can be obtained easily. A
players’ interpretations of the set of payoff functions numerical example is provided to illustrate the method.
(preferences) can be different from one another because
of misperceptions. Subscripts are used to describe such an
interpretation: Vij expresses player i's PV as interpreted 1. Introduction
Abstract
Game theory is a typical theoretical framework to deal
with decision situations with multiple conflicting players [1]. When a conflict is modeled as a game, the game model can be represented as follows.
G={N;S1,S2,L,Sn;v1,v2L,vn}. (1) where N={1,2,L,n} is a set of players; Si, i∈N is a set of strategies for each of the players; vi, i∈N is a set of payoff functions, which reflect the players’ preferences over the outcome space O=S1×S2×L×Sn.
If all the outcomes are ranked in order according to a player’s payoffs by writing the most preferred outcome on the left and least preferred on the right, then a preference vector (PV) is formed for the player and denoted by Vi. Consequently, a game can be represented by a set of PVs:
G={V1,V2,L,Vi,L,Vn}. (2)
by player j. When there is misperception, for instance, player j’s interpretation of player i's PV will be different from Vi, so that the game played by player j will be different from the one played by player i. A formal definition of an Lth-level hypergame model is given below [3]:
HL=H1L 1,H2L 1,L,HnL 1,L=1,2,3,L, (3) ( i,j∈N:HiL 1≠HLj 1).
{}
By using the above general formula, hypergame model on a given level can be written explicitly. In a first-level hypergame, at least one of the players is playing a different game from the others. Each player’s game is formed by a set of perceived PVs, except the one for the player himself. The first-level hypergame is formulated as
0000 (4) H1={H10,H2,L,H0j,L,Hn}( i,j∈N:Hi≠Hj)
T
where, H0j=(V1j,V2j,L,Vij,L,Vnj),j=1,2,L,n.
热门文档