储备池算法与动力系统分析研究进展
Reservoir computing and dynamical systems:a survey
高健 1颜子翔 1肖井华2
作者信息
- 1. 北京邮电大学理学院,100876,北京
- 2. 北京邮电大学理学院,100876,北京;北京邮电大学信息光子学与光通信国家重点实验室,100876,北京
- 折叠
摘要
储备池算法因为其简约的算法结构、灵活多变的算法实现方式,以及同时具有非线性和记忆性等特点,在时间序列和动力系统相关问题的研究中被广泛应用.同时,经过训练的储备池本身形成了一个复杂动力系统.以储备池为桥梁,基于动力系统理论的分析对于探索学习过程的智能现象起到了重要作用.本文对储备池算法的最新研究进展进行了回顾与展望,梳理出逐渐形成的"用于动力系统的储备池算法"与"储备池算法的动力系统基础"2个相辅相成的研究方向.本文将有助于研究人员加深对以储备池为代表的人工智能算法的理解,促进人工智能、复杂系统及统计物理相关学科跨学科研究和交流.
Abstract
Science exploration based on increasingly sophisticated artificial intelligence has witnessed great technological advances.Meanwhile,mathematical foundation of intelligent phenomena has emerged as a crucial scientific issue requiring urgent resolution.Of the numerous artificial intelligence algorithms,reservoir computing,due to its simple structure,flexible and diverse implementation,and characteristics of nonlinearity and memory storage,is widely applied in the research of time series and dynamical system-related problems.Trained reservoir forms a complex dynamical system.Serving as a bridge,analysis by dynamical system theory through the reservoir plays an important role in exploring intelligent phenomena in the learning process.The latest progress in reservoir computing algorithm is reviewed in this work,outlining two complementary research directions:reservoir computing for dynamical systems,dynamical system foundations of reservoir computing.The present work will deepen understanding of artificial intelligence algorithms,and promote interdisciplinary research and communication among artificial intelligence,complex systems,and statistical physics-related disciplines.
关键词
人工神经网络/储备池算法/动力系统/非线性动力学Key words
artificial neural network/reservoir computing/dynamical system/nonlinear dynamics引用本文复制引用
基金项目
国家自然科学基金(11775034)
中央高校基本科研业务费专项(2022RC26)
中央高校基本科研业务费专项(2019XD-A10)
出版年
2023
国家科技期刊平台
NSTL
维普
万方数据