“人工鱼”自学习方法研究

Study on Self-learning Method of“Artificial Fish”

“人工鱼”利用人工生命方法创作计算机动画。为了进一步提高动画角色的自适应能力，该文将人工神经网络方法和基于时间差分方法TD的强化学习相结合，建立了“人工鱼”的自学习模型，控制人工鱼的行为。该实验表明该“人工鱼”自学习系统可以使“人工鱼”成为更加自主和智能的角色。

In“artificial fish”, computer animation is created with artificial life. To improve self-adaption of animation role, methodology of ANN is integrated with reinforcement learning based on TD, and self-learning models of“artificial fish”are created to control behaviors of“artificial fish”. The experimental results have shown that self-learning system of“artificial fish”provides artificial life with more independence and intelligence.