Discovery of High-Level Behavior From Observation of Human Performance in a Strategic Game

This paper explores the issues faced in creating a sys-4 tem that can learn tactical human behavior merely by observing5 a human perform the behavior in a simulation. More specifically,6 this paper describes a technique based on fuzzy ARTMAP (FAM)7 neural networks to discover the criteria that cause a transition8 between contexts during a strategic game simulation. The ap-9 proach depends on existing context templates that can identify10 the high-level action of the human, given a description of the11 situation along with his action. The learning task then becomes the12 identification and representation of the context sequence executed13 by the human. In this paper, we present the FAM/Template-based14 Interpretation Learning Engine (FAMTILE). This system seeks to15 achieve this learning task by constructing rules that govern the16 context transitions made by the human. To evaluate FAMTILE, six17 test scenarios were developed to achieve three distinct evaluation18 goals: 1) to assess th...

• Type: Journal
• Year: 2008
• Where: TSMC
• Authors: Brian S. Stensrud, Avelino J. Gonzalez
• Keywords: Human, Interpretation Learning Engine, Strategic Game Simulation, TSMC 2008