基于进化算法的多智能体合作学习  

Multiagent cooperation learning based on an evolutional algorithm

作　　者：王云[1] 王俊[1] 韩伟[1] 

机构地区：[1]南京财经大学信息工程学院,江苏南京210046

出　　处：《山东大学学报（工学版）》2010年第6期8-11,共4页Journal of Shandong University（Engineering Science）

基　　金：国家自然科学基金资助项目(70802025)

摘　　要：强化学习的收敛速度随状态-动作空间的维数呈指数增长,因此在涉及大的状态空间时,强化学习算法的收敛速度非常慢以至不能满足应用需求。在许多应用环境中,若智能体之间存在合作关系,借助多个智能体进行分布式学习可以部分解决这一问题。利用进化算法,设计了智能体繁殖、消亡等操作,使得子代智能体能够继承父代智能体在状态空间的方向信息,从而更快地找到状态-动作空间的有效更新。仿真实验表明:算法比已有的强化学习方法具有更高的搜索效率和收敛速度。Reinforcement learning is not applicable concerning large state-actions, since that its convergence speed increases exponentially with the number of dimensions of state-action space. In many situations, this problem partially can be solved by utilizing a cooperation relationship among agents. An evolutional algorithm was put forward, which could rapidly find the effective updating of state-action pairs by the evolutionary operators such as reproduction as well as die out. Simulations proved that the algorithm performs was better than present multiagent cooperation learning algorithms.

关 键 词：多智能体系统 合作学习 进化算法 

分 类 号：TP181.1[自动化与计算机技术—控制理论与控制工程]