Speaker: Amiram (Ami) Moshaiov, School of Mechanical Engineering and Sagol School of Neuroscience, Tel Aviv University
"Conflict situations can be modelled and analyzed by a non-cooperative game-theoretic approach. In such situations the involved players may face not only the conflict with their opponent, but also their self-conflicting objectives. For example, in defense-offense problems, the attacker's objectives may be defined as reducing the time to reach a target while minimizing the number of casualties; these objectives are commonly conflicting. Such situations should be modeled as Multi-Objective Games (MOGs), which are also known as multi payoff, multi criteria or vector payoff games. The common approach to deal with MOGs is to assume that the objective preferences of the players are known a-priori. In such a case a utility function is used, which transforms the MOG into a surrogate single-objective game. However, players may have doubts when trying to a-priori make a rational decision on their objective preferences.
This presentation deals with non-cooperative MOGs in a non-traditional way, which is inspired by Pareto-based multi-objective optimization. The zero-sum MOG model, which is considered here, involves two players that are undecided about their objective preferences. This unique problem definition allows finding a set of rationalizable strategies for each player. Consequently, the players can decide on their preferred strategies based on a comprehensive analysis of their alternative rationalizable strategies. To illustrate the idea, a MOG between competing traveling salespersons is introduced and an associated simple case study is presented and analyzed. First, we show how the sets of rationalizable strategies are defined and solved based on worst-case domination relation among sets of performance vectors. Second, we demonstrate how the set of rationalizable strategies can be used, via multi-criteria decision-making techniques, to select a preferred strategy."