To create DDA by the approach of ANN from UCT-created data
Xinyu Li, Suoju He, Yue Dong, Qing Fei Liu, Xiao Liu, Yiwen Fu, Zhiyuan Shi, Wan Huang · 2010
Dynamic Difficulty Adjustment (DDA) can adjust game difficulty level dynamically; so it generates a tailor-made experience for each gamer. If a game is too easy, the gamer will feel bored; if it is too hard, the gamer will become frustrated. DDA is a mechanism to overcome this dilemma and augment the entertainment of a game by dynamically adjusting the parameters, scenarios and behaviors in the game in real-time based on the gamer's personal ability. We use Upper Confidence bound for Trees (UCT) to create the training data, and then train the Artificial Neural Networks (ANN) off-line with that data. Finally, we derive DDA from ANN approach. In this paper, the prey and predator game genre of Pac-Man is utilized as a test-bed, the procedure of training ANN is shown, and the feasibility of applying DDA to game artificial intelligence (AI) development is demonstrated.