摘要
本文研究具有人机交互能力的强化学习算法。通过人机交互给出操作者对学习结果的性能评价,智能体系统能获得当前状态与目标状态距离的度量,有效地结合操作者的先验知识和专业知识,从而使智能体在状态空间中能进行更有效的搜索,简化复杂任务的学习过程。以猜数字游戏为例,利用提出的学习框架训练智能体具有猜数字的能力。实验结果表明,结合人机交互的强化学习算法大大提高了学习效率。加快了学习过程的收敛速度。
In this paper, a reinforcement learning algorithm based on human-computer interaction is proposed. This interactive learning system can benefit from measurements of the distance between current state and goal state via operator's professional knowledge. Thus learning procedure is expected to be more efficient. A guess-number task is explored to evaluate the proposed learning system. Experimental result shows that the learning efficiency and convergence rate are both increased compared with normal reinforcement learning method.
出处
《模式识别与人工智能》
EI
CSCD
北大核心
2003年第3期363-369,共7页
Pattern Recognition and Artificial Intelligence
基金
国家自然科学基金(No.60275042)
