深度多代理强化学习的迁移学习框架

IF 15.3 1区计算机科学 Q1 AUTOMATION & CONTROL SYSTEMS

Ieee-Caa Journal of Automatica Sinica Pub Date : 2024-10-08 DOI:10.1109/JAS.2023.124173

Yi Liu;Xiang Wu;Yuming Bo;Jiacun Wang;Lifeng Ma

{"title":"深度多代理强化学习的迁移学习框架","authors":"Yi Liu;Xiang Wu;Yuming Bo;Jiacun Wang;Lifeng Ma","doi":"10.1109/JAS.2023.124173","DOIUrl":null,"url":null,"abstract":"Dear Editor, This letter presents a new transfer learning framework for the deep multi-agent reinforcement learning (DMARL) to reduce the convergence difficulty and training time when applying DMARL to a new scenario [1], [2]. The proposed transfer learning framework includes the design of neural network architecture, curriculum transfer learning (CTL) and strategy distillation. Experimental results demonstrate that our framework enables DMARL models to converge faster while improving the final performance.","PeriodicalId":54230,"journal":{"name":"Ieee-Caa Journal of Automatica Sinica","volume":"11 11","pages":"2346-2348"},"PeriodicalIF":15.3000,"publicationDate":"2024-10-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10707687","citationCount":"0","resultStr":"{\"title\":\"A Transfer Learning Framework for Deep Multi-Agent Reinforcement Learning\",\"authors\":\"Yi Liu;Xiang Wu;Yuming Bo;Jiacun Wang;Lifeng Ma\",\"doi\":\"10.1109/JAS.2023.124173\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Dear Editor, This letter presents a new transfer learning framework for the deep multi-agent reinforcement learning (DMARL) to reduce the convergence difficulty and training time when applying DMARL to a new scenario [1], [2]. The proposed transfer learning framework includes the design of neural network architecture, curriculum transfer learning (CTL) and strategy distillation. Experimental results demonstrate that our framework enables DMARL models to converge faster while improving the final performance.\",\"PeriodicalId\":54230,\"journal\":{\"name\":\"Ieee-Caa Journal of Automatica Sinica\",\"volume\":\"11 11\",\"pages\":\"2346-2348\"},\"PeriodicalIF\":15.3000,\"publicationDate\":\"2024-10-08\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10707687\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Ieee-Caa Journal of Automatica Sinica\",\"FirstCategoryId\":\"94\",\"ListUrlMain\":\"https://ieeexplore.ieee.org/document/10707687/\",\"RegionNum\":1,\"RegionCategory\":\"计算机科学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q1\",\"JCRName\":\"AUTOMATION & CONTROL SYSTEMS\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Ieee-Caa Journal of Automatica Sinica","FirstCategoryId":"94","ListUrlMain":"https://ieeexplore.ieee.org/document/10707687/","RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"AUTOMATION & CONTROL SYSTEMS","Score":null,"Total":0}

引用次数: 0

摘要

亲爱的编辑，这封信为深度多代理强化学习（DMARL）提出了一个新的迁移学习框架，以减少将DMARL应用于新场景时的收敛难度和训练时间[1], [2]。所提出的迁移学习框架包括神经网络架构设计、课程迁移学习（CTL）和策略提炼。实验结果表明，我们的框架能使 DMARL 模型更快地收敛，同时提高最终性能。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

查看原文本刊更多论文

A Transfer Learning Framework for Deep Multi-Agent Reinforcement Learning

Dear Editor, This letter presents a new transfer learning framework for the deep multi-agent reinforcement learning (DMARL) to reduce the convergence difficulty and training time when applying DMARL to a new scenario [1], [2]. The proposed transfer learning framework includes the design of neural network architecture, curriculum transfer learning (CTL) and strategy distillation. Experimental results demonstrate that our framework enables DMARL models to converge faster while improving the final performance.

求助全文

通过发布文献求助，成功后即可免费获取论文全文。去求助

来源期刊

Ieee-Caa Journal of Automatica Sinica Engineering-Control and Systems Engineering

CiteScore

23.50

自引率

11.00%

发文量

880

期刊介绍： The IEEE/CAA Journal of Automatica Sinica is a reputable journal that publishes high-quality papers in English on original theoretical/experimental research and development in the field of automation. The journal covers a wide range of topics including automatic control, artificial intelligence and intelligent control, systems theory and engineering, pattern recognition and intelligent systems, automation engineering and applications, information processing and information systems, network-based automation, robotics, sensing and measurement, and navigation, guidance, and control. Additionally, the journal is abstracted/indexed in several prominent databases including SCIE (Science Citation Index Expanded), EI (Engineering Index), Inspec, Scopus, SCImago, DBLP, CNKI (China National Knowledge Infrastructure), CSCD (Chinese Science Citation Database), and IEEE Xplore.