Data mining considerations for knowledge acquisition in real time strategy games
Gabriel Iuhasz, Victor Ion Munteanu, Viorel Negru · 2013
Adaptive Game AI has been one of the key topics being researched in the field of academic game AI research. In this paper we present a comparison of several domain independent machine learning methods with the aid of which we extract expert knowledge from game logs. Each game log is represented as a feature vector that encodes cardinality and timing for player actions. We compare a wide variety of classification methods and highlight which ones are best for deployment for an adaptive game AI systems.