Learning by Observing: Case-Based Decision Making in Complex Strategy Games

Darko Obradovic, Armin Stahl
Proceedings of the 31th Annual German Conference on Artificial Intelligence, Kaiserslautern, Germany, Springer, 9/2008

Abstract:

There is a growing research interest in the design of competitive and adaptive Game AI for complex computer strategy games. In this paper, we present a novel approach for developing intelligent bots, which is based on the idea to observe successful human players and to learn from their individual decisions and strategies. These decisions are then reused by a bot in similar situations, resulting in a flexible and realistic strategic behaviour with low development and knowledge acquisition costs. Using Case-Based Reasoning (CBR) techniques, we implement this principle in the Cyborg system and achieve to outperform scripted opponents in a challenging multiplayer scenario.

Files:

  http://www.springerlink.com/content/f510543717687467/

BibTex:

@inproceedings{ OBRA2008,
	Title = {Learning by Observing: Case-Based Decision Making in Complex Strategy Games},
	Author = {Darko Obradovic and Armin Stahl},
	BookTitle = {Proceedings of the 31th Annual German Conference on Artificial Intelligence},
	Month = {9},
	Year = {2008},
	Publisher = {Springer}
}

     
Last modified:: 30.08.2016