Overcoming intermittent instability in reinforcement learning via gradient norm preservation

IF 8.1 1区 计算机科学 0 COMPUTER SCIENCE, INFORMATION SYSTEMS Information Sciences Pub Date : 2025-03-13 DOI:10.1016/j.ins.2025.122081
Jonghyeok Park , Jongsoo Lee , Jangwon Kim , Soohee Han
{"title":"Overcoming intermittent instability in reinforcement learning via gradient norm preservation","authors":"Jonghyeok Park ,&nbsp;Jongsoo Lee ,&nbsp;Jangwon Kim ,&nbsp;Soohee Han","doi":"10.1016/j.ins.2025.122081","DOIUrl":null,"url":null,"abstract":"<div><div>Instability is a critical challenge in online reinforcement learning (RL), particularly in robotics. While existing RL methods primarily address this intractable instability through algorithmic modifications or training strategies, the role of optimization techniques remains largely underexplored. This paper investigates intermittent instability in RL training, which hinders accurate value learning, and proposes a novel optimization approach: Gradient Norm Preservation (GNP). Our analysis identifies irregular gradient spikes, caused by high-reward data during exploration, as a key source of instability. These spikes are mathematically quantified, and the optimizer's learning rate is dynamically adjusted to preserve initial gradient norms, mitigating their impact on value learning. Experiments across diverse environments demonstrate that integrating GNP into RL algorithms significantly improves stability, with notable gains in training performance across several environments.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"709 ","pages":"Article 122081"},"PeriodicalIF":8.1000,"publicationDate":"2025-03-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Information Sciences","FirstCategoryId":"94","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0020025525002130","RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"0","JCRName":"COMPUTER SCIENCE, INFORMATION SYSTEMS","Score":null,"Total":0}
引用次数: 0

Abstract

Instability is a critical challenge in online reinforcement learning (RL), particularly in robotics. While existing RL methods primarily address this intractable instability through algorithmic modifications or training strategies, the role of optimization techniques remains largely underexplored. This paper investigates intermittent instability in RL training, which hinders accurate value learning, and proposes a novel optimization approach: Gradient Norm Preservation (GNP). Our analysis identifies irregular gradient spikes, caused by high-reward data during exploration, as a key source of instability. These spikes are mathematically quantified, and the optimizer's learning rate is dynamically adjusted to preserve initial gradient norms, mitigating their impact on value learning. Experiments across diverse environments demonstrate that integrating GNP into RL algorithms significantly improves stability, with notable gains in training performance across several environments.

Abstract Image

查看原文
分享 分享
微信好友 朋友圈 QQ好友 复制链接
本刊更多论文
不稳定性是在线强化学习(RL),尤其是机器人技术中的一个关键挑战。现有的 RL 方法主要通过算法修改或训练策略来解决这种难以解决的不稳定性问题,而优化技术的作用在很大程度上仍未得到充分发挥。本文研究了 RL 训练中阻碍精确数值学习的间歇性不稳定性,并提出了一种新颖的优化方法:梯度规范保护(GNP)。我们的分析发现,在探索过程中由高回报数据引起的不规则梯度尖峰是不稳定性的主要来源。我们对这些峰值进行了数学量化,并动态调整优化器的学习率,以保持初始梯度规范,减轻其对价值学习的影响。不同环境下的实验证明,将 GNP 集成到 RL 算法中能显著提高稳定性,在多个环境下的训练性能都有明显提高。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 去求助
来源期刊
Information Sciences
Information Sciences 工程技术-计算机:信息系统
CiteScore
14.00
自引率
17.30%
发文量
1322
审稿时长
10.4 months
期刊介绍: Informatics and Computer Science Intelligent Systems Applications is an esteemed international journal that focuses on publishing original and creative research findings in the field of information sciences. We also feature a limited number of timely tutorial and surveying contributions. Our journal aims to cater to a diverse audience, including researchers, developers, managers, strategic planners, graduate students, and anyone interested in staying up-to-date with cutting-edge research in information science, knowledge engineering, and intelligent systems. While readers are expected to share a common interest in information science, they come from varying backgrounds such as engineering, mathematics, statistics, physics, computer science, cell biology, molecular biology, management science, cognitive science, neurobiology, behavioral sciences, and biochemistry.
期刊最新文献
Approaches to attribute reduction of metric-fuzzy decision systems based on information theory Overcoming intermittent instability in reinforcement learning via gradient norm preservation Random walk on point clouds for feature detection Toward rough set based insightful reasoning in intelligent systems Knowledge&Social-based collaborative method with contrastive graph structure learning for explainable recommendation
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
现在去查看 取消
×
提示
确定
0
微信
客服QQ
Book学术公众号 扫码关注我们
反馈
×
意见反馈
请填写您的意见或建议
请填写您的手机或邮箱
已复制链接
已复制链接
快去分享给好友吧!
我知道了
×
扫码分享
扫码分享
Book学术官方微信
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术
文献互助 智能选刊 最新文献 互助须知 联系我们:info@booksci.cn
Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。
Copyright © 2023 Book学术 All rights reserved.
ghs 京公网安备 11010802042870号 京ICP备2023020795号-1