A Multiple-Goal Sarsa(λ) Algorithm Based on Lost Reward of Greatest Mass

LIU Quan; LI Jin; FU Qi-ming; CUI Zhi-ming; FU Yu-chen

doi:10.3969/j.issn.0372-2112.2013.08.003

您当前的位置：

首页 >

文章列表页 >

A Multiple-Goal Sarsa(λ) Algorithm Based on Lost Reward of Greatest Mass

更新时间：2025-07-16

- A Multiple-Goal Sarsa(λ) Algorithm Based on Lost Reward of Greatest Mass
- Acta Electronica Sinica Vol. 41, Issue 8, Pages: 1469-1473(2013)
- 作者机构：
  
  1. 苏州大学计算机与科学学院,江苏,苏州,215000
  2. 符号计算与知识工程教育部重点实验室(吉林大学),吉林,长春,130012
  3. 苏州大学计算机与科学学院,江苏,苏州,215000
  4. 符号计算与知识工程教育部重点实验室(吉林大学),吉林,长春,130012
- 作者简介：
- 基金信息：
- DOI：10.3969/j.issn.0372-2112.2013.08.003
  CLC： TP181
- Published：2013
- 稿件说明：
移动端阅览
LIU Quan, LI Jin, FU Qi-ming, et al. A Multiple-Goal Sarsa(λ) Algorithm Based on Lost Reward of Greatest Mass[J]. Acta Electronica Sinica, 2013, 41(8): 1469-1473.
DOI：

LIU Quan, LI Jin, FU Qi-ming, et al. A Multiple-Goal Sarsa(λ) Algorithm Based on Lost Reward of Greatest Mass[J]. Acta Electronica Sinica, 2013, 41(8): 1469-1473. DOI： 10.3969/j.issn.0372-2112.2013.08.003.

摘要

针对RoboCup这一典型的多目标强化学习问题

提出一种基于最大集合期望损失的多目标强化学习算法LRGM-Sarsa(

)算法.该算法预估各个目标的最大集合期望损失

在平衡各个目标的前提下选择最佳联合动作以产生最优联合策略.在单个目标训练的过程中

采用基于改进MSBR误差函数的Sarsa(

)算法

并对动作选择概率函数和步长参数进行优化

解决了强化学习在使用非线性函数泛化时

算法不稳定、不收敛的问题.将该算法应用到RoboCup射门局部策略训练中

取得了较好的效果

表明该学习算法的有效性.

Abstract

For solving the multiple-goal problem in RoboCup

a novel multiple-goal Reinforcement Learning algorithm

named LRGM-Sarsa(

)

is proposed.The algorithm estimates the lost reward of the greatest mass of every sub goal and trades off the long term reward of the sub goals to get a composite policy.In the single learning module

B error function

which is based on MSBR error function is proposed.B error function has guaranteed the convergence of the value prediction with the non-linear function approximation.The probability funciton of selecting actions and the parameter

are also improved with respect to B error function.This algorithm is applied to the training of shooting in Robocup 2D.The experimental results show that the pro

posed algorithm is more stable and converges faster.

关键词

Keywords

references

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

CS-ROMF:Efficient Community Search Based on Graph Combinatorial Optimization

Knowledge-Based and Data-Driven Integrating Design Methodology for Air Combat Strategy in Multi-Opponent Adversarial Game

Reinforcement Learning Based Tuning-free Plug-and-Play Image Reconstruction Method for Single Photon Imaging

Optimal Directed Control of Discrete Event Systems Based on Reinforcement Learning

Autoimmune Dynamic Attack Generation Method Based on Reinforcement Learning

Related Author

ZHANG An-ran

WANG Xing-fen

ZHAO Yu-han

LI Li-bo

FENG Jin-yuan

CHEN Min

LI Jun-ying

CHEN Jia-le

Related Institution

Beijing Information Science and Technology University

Hong Kong Baptist University, Hongkong

Institute of Automation, Chinese Academy of Sciences

School of Artificial Intelligence, University of Chinese Academy of Sciences

School of Computer Science, Nanjing University of Information Science & Technology

⁰