论文信息 - A Demonstration of Agent Learning with Action-Dependent Learning Rates in Computer Role-Playing Games

A Demonstration of Agent Learning with Action-Dependent Learning Rates in Computer Role-Playing Games

We demonstrate combat scenarios between two NPCs in the Neverwinter Nights (NWN) game in which an NPC uses a new learning algorithm ALeRT (Action-dependent Learning Rates with Trends) and the other NPC uses a static strategy (NWN default and optimal) or a dynamic strategy (dynamic scripting). We implemented the ALeRT algorithm in NWScript, a scripting language used by NWN, with the goal to improve the behaviours of game agents. We show how our agent learns and adapts to changes in the environment.

Duane Szafron | Maria Cutumisu | D. Szafron | M. Cutumisu

[1] Eric O. Postma,et al. On-line Adaptation of Game Opponent AI with Dynamic Scripting , 2004, Int. J. Intell. Games Simul..

[2] Duane Szafron,et al. Agent Learning using Action-Dependent Learning Rates in Computer Role-Playing Games , 2008, AIIDE.

[3] Richard S. Sutton,et al. Adapting Bias by Gradient Descent: An Incremental Version of Delta-Bar-Delta , 1992, AAAI.

[4] Richard S. Sutton,et al. Introduction to Reinforcement Learning , 1998 .

[5] Richard S. Sutton,et al. Reinforcement Learning: An Introduction , 1998, IEEE Trans. Neural Networks.

[6] Manuela M. Veloso,et al. Rational and Convergent Learning in Stochastic Games , 2001, IJCAI.