摘要： 提出了一种新的网格资源预测方法。该算法基于TD学习,不需建立模型,直接从预测环境获得反馈信息,再从反馈信息中学习相应的规则。给出了该算法Java语言的一种实现,并用该系统对电信的接入主机的性能进行了预测试验。结果表明,该预测算法是有效的。同时给出了学习速度参数和历史信息相关度参数对预测的影响。

关键词: 强化学习, 瞬时差分学习, 网格, 资源预测

Abstract: This paper presents a novel prediction algorithm,which is based on temporal difference learning.It can learn directly from raw experience without a model of the environment.An implemention is developed with Java language,and it is tested on the host running in the Internet.The experiment results show that this method works remarkably well.An analysis about how the parameters of learning rate and history correlativity impact the prediction is given.

中图分类号: 

• TP393