90 likes | 103 Views
Generating War Game Strategies Using A Genetic Algorithm 1. Advisor : Dr. Hsu Presenter : Jia-Hao Yang Author :Timothy E. Revello, Robert McCartney. Outline. Motivation Objective Introduction Method (GA) Experiment Conclusion. Motivation.
E N D
Generating War Game Strategies Using A Genetic Algorithm1 Advisor : Dr. Hsu Presenter : Jia-Hao Yang Author :Timothy E. Revello, Robert McCartney IEEE EC
Outline • Motivation • Objective • Introduction • Method (GA) • Experiment • Conclusion IEEE EC
Motivation • Unlike most games which have fixed rules, the rules for war games can contain uncertainty. • Although there are many approach about game AI have been developed, but all based on the game structure and rules being fixed. IEEE EC
Objective • In this paper we explore the use of genetic algorithms for generating war game strategies. IEEE EC
Introduction • Gaming is used extensively by the military and business for decision making…etc. • So game play has been a topic of research in computational science for some time. • EX : tic-tac-toe, eight tile puzzles, Checkers and chess, war game. • One of the challenges currently faced is to better address uncertainty. • Variables working. ex: what moves the opposition will make • Rules of the game themselves. ex: what is required to win are generally not known. IEEE EC
Method • The use of a GA for generating war game strategies using a naval blockade scenario as a tool. • Naval forces are modeled using a time step simulation. Blue win criteria IEEE EC
Criterion method elitism Count method Criterion method random Count method Method • GA configuration : Include 396 bits long, consisting of 12 groups of 33 bits/per ship group, which include the type, number, and movement of ship groups encoding Include 30 individuals population Fitness function 5000 Parent selection Crossover 1.0 0.001 Mutation IEEE EC
Experiment • The result showed that the most robust configuration was the use of elitism with the count method. IEEE EC
Conclusion • During this investigation we gave demonstrated that a GA can be used to generate strategies for a war game in which the rules are uncertain. IEEE EC