强化学习的模型、算法及应用  

Reinforcement Learning Model,Algorithms and Its Application

作　　者：战忠丽[1] 王强[1] 陈显亭[1] 

机构地区：[1]吉林电子信息职业技术学院计算机系,吉林吉林132021

出　　处：《电子科技》2011年第1期47-49,共3页Electronic Science and Technology

摘　　要：强化学习不需要具有先验知识,通过试错与环境交互获得策略的改进,具有自学习和在线学习能力,是构造智能体的核心技术之一。文中首先综述了强化学习模型和基本原理,然后介绍了强化学习的主要算法,包括Sarsa算法、TD算法、Q-学习算法及函数估计算法,最后介绍了强化学习的应用情况及未来研究方向。Reinforcement Learning does not need prior knowledge, and it gets optional policy through trial and error, its capacity of self-improving and online learning is one of the basic technologies of intelligent agent. In the paper, we firstly introduce the model and foundation of RL, then, we deeply discuss the main algorithms of RL, including Sarsa, temporal difference, Q-learning and function approximation, finally, we briefly introduce some applications of RL and some future research direction.

关 键 词：强化学习 Sarsa算法 瞬时差分算法 Q-学习算法 函数估计 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]