节点文献
“人工鱼”自学习方法研究
Study on Self-learning Method of“Artificial Fish”
【摘要】 “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。
【Abstract】 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.
【关键词】 人工生命;
自学习;
强化学习;
人工神经网络;
【Key words】 Artificial life; Self-learning; Reinforcement learning; Artificial neural nets(ANN);
【Key words】 Artificial life; Self-learning; Reinforcement learning; Artificial neural nets(ANN);
【基金】 国家自然科学基金资助项目(60075012);清华大学智能技术与系统国家重点实验室开放课题基金资助项目(0109)
- 【文献出处】 计算机工程 ,Computer Engineering , 编辑部邮箱 ,2004年06期
- 【分类号】TP391.41
- 【被引频次】19
- 【下载频次】295