摘要
传统的学习方法在调度网络共享资源时收敛速度较慢,各个节点上的负载能量很小。深度强化学习将深度学习和强化学习结合在一起,提升决策能力,直接控制资源数据,思维方式与人类的思维方式相似,智能性极高。结合深度学习和强化学习研究了一种网络共享资源调度方法,该方法的总体框架分为四步运行,首先利用网络节点得到资源数据;然后使用深度学习算法训练资源数据;接着以强化学习的方式将网络资源数据合理分配;最后通过参数校正完成网络共享资源的合理调度。与传统调度方法进行实验对比,结果表明,基于深度强化学习的网络共享资源智能调度方法能够有效提高收敛速度,增加各个网络节点上的负载能量,具有很好的调度性能。
Traditional learning methods converge slowly in scheduling network shared resources,and the load energy on each node is very small.Deep reinforcement learning combines in-depth learning with reinforcement learning to improve decision-making ability and directly control resource data.Its thinking mode is similar to human thinking mode and has high intelligence.In this paper,a network shared resource scheduling method based on in-depth learning and reinforcement learning is studied.The overall framework of this method is divided into four steps.Firstly,resource data is obtained by using network nodes.Secondly,resource data is trained by using in-depth learning algorithm,and network resource data is allocated reasonably by reinforcement learning.Finally,the network resource data is distributed reasonably.The over parameter correction completes the rational scheduling of network shared resources.Compared with the traditional scheduling method,the results show that the intelligent scheduling method based on deep reinforcement learning can effectively improve the convergence rate,increase the load energy on each network node,and has good scheduling performance.
作者
何杨
肖基毅
HE Yang;XIAO Jiyi(School of Computer Science,Nanhua University,Hengyang 421000,China)
出处
《自动化与仪器仪表》
2019年第6期80-82,90,共4页
Automation & Instrumentation
关键词
深度强化
强化学习
网络共享
资源调度
智能调度
deep reinforcement
reinforcement learning
network sharing
resource scheduling
intelligent scheduling
