A chaotic behavior decision algorithm based on self-generating neural network for computer games

The artificial intelligence of computer games is constantly in need of improvement to meet the increasing demands of game players. This paper proposes a kind of chaotic behavior decision algorithm, which integrates fuzzy logic and based on self-generating neural network. In computer games, this method is applied to achieve the goal of improving the non-player characterspsila (NPCs) behavior ability, therefore contributing to enhance computer games intelligence level based on human knowledge and experience. As one of novel developed neural networks for classify and prediction, SGNN has the features of simplicity in network design, fast learning and automatic organizing ability. Incorporated with fuzzy method, the chaotic behavior decision algorithm includes two parts: offline behavior rule extraction and online behavior decision. The method proposed by this paper in game intelligence is flexible and adaptive to generic games. It has more powerful ability in behavior selection. The corresponding application will be illustrated in detail through implementation on special military game.