Adaptive Dynamic Programming: An Introduction

IF 7.7 2区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

IEEE Computational Intelligence Magazine Pub Date : 2009-05-01 DOI:10.1109/MCI.2009.932261

Fei-Yue Wang, Huaguang Zhang, Derong Liu

引用次数: 761

Abstract

In this article, we introduce some recent research trends within the field of adaptive/approximate dynamic programming (ADP), including the variations on the structure of ADP schemes, the development of ADP algorithms and applications of ADP schemes. For ADP algorithms, the point of focus is that iterative algorithms of ADP can be sorted into two classes: one class is the iterative algorithm with initial stable policy; the other is the one without the requirement of initial stable policy. It is generally believed that the latter one has less computation at the cost of missing the guarantee of system stability during iteration process. In addition, many recent papers have provided convergence analysis associated with the algorithms developed. Furthermore, we point out some topics for future studies.

查看原文本刊更多论文

自适应动态规划:导论

本文介绍了自适应/近似动态规划(ADP)领域的一些最新研究动向，包括ADP方案结构的变化、ADP算法的发展和ADP方案的应用。对于ADP算法，重点在于ADP的迭代算法可以分为两类:一类是具有初始稳定策略的迭代算法;另一种是不要求初始稳定政策的。一般认为，后者的计算量较少，但代价是在迭代过程中失去了对系统稳定性的保证。此外，最近的许多论文都提供了与所开发的算法相关的收敛分析。在此基础上，提出了今后的研究方向。

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分钟内获得全文求助全文

来源期刊

IEEE Computational Intelligence Magazine 工程技术-计算机：人工智能

CiteScore

14.50

自引率

3.30%

发文量

审稿时长

>12 weeks

期刊介绍： The IEEE Computational Intelligence Magazine (CIM) is a peer-reviewed publication that focuses on computational intelligence design and applications. It aims to present emerging novel discoveries, important insights, and tutorial surveys in this field. CIM serves as a means of communication between the governing body and members of the IEEE Computational Intelligence Society (IEEE/CIS). The magazine encourages authors to submit papers on various topics, including applications oriented developments, successful industrial implementations, design tools, technology reviews, computational intelligence education, and applied research. Contributions should contain novel and previously unpublished material. The novelty can be in the form of original concepts, results, techniques, observations, hardware/software implementations, or applications. Syntheses or new insights into previously reported research are also welcomed. Surveys and expository submissions are highly encouraged. Generally, material that has been previously copyrighted, published, or accepted for publication will not be considered for publication in CIM. However, prior preliminary or abbreviated publication of the material will not necessarily preclude publication in this journal.