AccScience Publishing / NSCE / Online First / DOI: 10.36922/NSCE025290003
Submit to NSCE
Cite this article
9
Download
91
Views
Journal Browser
Volume | Year
Issue
Search
Forthcoming Issue
Current Issue
View All
News and Announcements
View All
RESEARCH ARTICLE

Predicting chaotic system behavior using machine learning techniques

Huaiyuan Rao1* ,  Yichen Zhao1 ,  Hsuan-Pin Chen1
Show Less
1 School of Electrical and Computer Engineering, Georgia Institute of Technology, Atlanta, Georgia, United States of America
NSCE, 025290003 https://doi.org/10.36922/NSCE025290003
Received: 18 July 2025 | Revised: 7 August 2025 | Accepted: 12 August 2025 | Published online: 26 August 2025
© 2025 by the Author(s). This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution -Noncommercial 4.0 International License (CC-by the license) ( https://creativecommons.org/licenses/by-nc/4.0/ )
Download PDF
XML
Cite
Abstract

Recently, the superior performance of machine learning approaches over classical forecasting models for complex time series analysis has been demonstrated in various domains. However, predicting chaotic time series continues to be a major challenge due to their inherent complexity. This paper investigates the accuracy, efficiency, and robustness of three recurrent neural network architectures: (i) next generation reservoir computing, (ii) reservoir computing, (iii) long short-term memory for the task of chaotic systems prediction. Four canonical chaotic systems, namely the Lorenz, Rössler, Chen, and Qi systems, are used for comparing these three methods. Numerical results demonstrate that the next generation reservoir computing is more computationally efficient and offers greater potential for predicting long-term chaotic time series.

Keywords
Time series prediction
Chaotic time series
Recurrent neural networks
Funding
None.
Conflict of interest
The authors declare they have no competing interests.
References
  1. Roondiwala M, Patel H, Varma S, et al. Predicting stock pricesusing LSTM. Int J Sci Res. 2017;6(4):1754-1756.

 

  1. Sheth D, Shah M. Predicting stock market using machinelearning: best and accurate way to know future stock prices.Int J Syst Assur Eng Manag. 2023;14(1):1-18.

 

  1. Price I, Sanchez-Gonzalez A, Alet F, et al. Probabilisticweather forecasting with machine learning. Nature.2025;637(8044):84-90.

 

  1. Fan J, Zhang K, Huang Y, Zhu Y, Chen B.Parallel spatiotemporal attention-based TCN formultivariate time series prediction. Neural Comput Appl.2023;35(18):13109-13118.

 

  1. Alsharef A, Aggarwal K, Sonia, Kumar M, Mishra A. Reviewof ML and AutoML solutions to forecast time-series data. ArchComput Methods Eng. 2022;29(7):5297-5311.

 

  1. Li G, Jung JJ. Deep learning for anomaly detection inmultivariate time series: approaches, applications, andchallenges. Inf Fus. 2023;91:93-102.

 

  1. Torres JF, Hadjout D, Sebaa A, Martínez-Álvarez F, TroncosoA. Deep learning for time series forecasting: a survey. BigData. 2021;9(1):3-21.

 

  1. Kong X, Chen Z, Liu W, et al. Deep learning for time seriesforecasting: a survey. Int J Mach Learn Cybern. 2025:1-34.

 

  1. Salman D, Direkoglu C, Kusaf M, Fahrioglu M. Hybrid deeplearning models for time series forecasting of solar power.Neural Comput Appl. 2024;36(16):9095-9112.

 

  1. Han M, Xi J, Xu S, Yin FL. Prediction of chaotic time seriesbased on the recurrent predictor neural network. IEEE TransSignal Process. 2004;52(12):3409-3416.

 

  1. Chandra R, Zhang M. Cooperative coevolution of Elmanrecurrent neural networks for chaotic time series prediction.Neurocomputing. 2012;86:116-123.

 

  1. Dudukcu HV, Taskiran M, Taskiran ZGC, Yildirim T.Temporal convolutional networks with RNN approachfor chaotic time series prediction. Appl Soft Comput.2023;133:109945.

 

  1. Shi H, Xu M, Li R. Deep learning for household loadforecasting—a novel pooling deep RNN. IEEE Trans SmartGrid. 2017;9(5):5271-5280.

 

  1. Ahmed NK, Atiya AF, Gayar NE, El-Shishiny H. An empiricalcomparison of machine learning models for time seriesforecasting. Econometric Rev. 2010;29(5-6):594-621.

 

  1. Wang L, Dai L. Chaotic time series prediction ofmulti-dimensional nonlinear system based on bidirectionalLSTM model. Adv Theory Simul. 2023;6(8):2300148.

 

  1. Yanan G, Xiaoqun C, Bainian L, Kecheng P. Chaotic timeseries prediction using LSTM with CEEMDAN. J Phys ConfSer. 2020;1617:012094.

 

  1. Langeroudi MK, Yamaghani MR, Khodaparast S. FD-LSTM:a fuzzy LSTM model for chaotic time-series prediction. IEEEIntell Syst. 2022;37(4):70-78.

 

  1. Sangiorgio M, Dercole F. Robustness of LSTM neuralnetworks for multi-step forecasting of chaotic time series.Chaos Solitons Fractals. 2020;139:110045.

 

  1. Pathak J, Hunt B, Girvan M, Lu Z, Ott E. Model-freeprediction of large spatiotemporally chaotic systems fromdata: a reservoir computing approach. Phys Rev Lett.2018;120(2):024102.

 

  1. Rafayelyan M, Dong J, Tan Y, Krzakala F, Gigan S.Large-scale optical reservoir computing for spatiotemporalchaotic systems prediction. Phys Rev X. 2020;10(4):041037.

 

  1. Carleo G, Cirac I, Cranmer K, Daudet L, Schuld M, TishbyN, Vogt-Maranto L, Zdeborova L. Machine learning and thephysical sciences. Rev Mod Phys. 2019;91(4):045002.

 

  1. Yan M, Huang C, Bienstman P, Tino P, LinW, Sun J. Emergingopportunities and challenges for the future of reservoircomputing. Nat Commun. 2024;15(1):2056.

 

  1. Vlachas PR, Pathak J, Hunt BR, et al. Backpropagationalgorithms and reservoir computing in recurrent neuralnetworks for the forecasting of complex spatiotemporaldynamics. Neural Netw. 2020;126:191-217.

 

  1. Bollt E. On explaining the surprising success of reservoircomputing forecaster of chaos? The universal machinelearning dynamical system with contrast to VAR and DMD.Chaos. 2021;31(1):013123.

 

  1. Gauthier DJ, Bollt E, Griffith A, Barbosa WA. Next generationreservoir computing. Nat Commun. 2021;12(1):1-8.

 

  1. Madondo M, Gibbons T. Learning and modeling chaos usingLSTM recurrent neural networks. Proceedings of the MidwestInstruction and Computing Symposium, Duluth, Minnesota;2018:6-7.

 

  1. Li Q, Lin R-C. A new approach for chaotic time seriesprediction using recurrent neural network. Math Probl Eng.2016;2016:3542898.

 

  1. Wang R, Peng C, Gao J, Gao Z, Jiang H. A dilated convolutionnetwork-based LSTM model for multi-step prediction ofchaotic time series. Comput Appl Math. 2020;39(1):30.

 

  1. Sangiorgio M. Deep learning in multi-step forecastingof chaotic dynamics. In: Special Topics in InformationTechnology. Cham: Springer International Publishing;2022:3-14.

 

  1. Sun Y, Zhang L, Yao M. Chaotic time series prediction ofnonlinear systems based on various neural network models.Chaos Solitons Fractals. 2023;175:113971.

 

  1. Lu X-Q, Tian J, Liao Q, et al. CNN-LSTM based incrementalattention mechanism enabled phase-space reconstructionfor chaotic time series prediction. J Electron Sci Technol.2024;22:100256.

 

  1. Serrano-Perez JDJ, Fernandez-Anaya G, Carrillo-MorenoS, Yu W. New results for prediction of chaotic systemsusing deep recurrent neural networks. Neural Process Lett.2021;53(2):1579-1596.

 

  1. Shi L, Wang H, Wang S, Du R, Qu S-X. Predicting nonsmoothchaotic dynamics by reservoir computing. Phys Rev E.2024;109(1):014214.

 

  1. Luo H, Du Y, Fan H, Wang X, Guo J, Wang X. Reconstructingbifurcation diagrams of chaotic circuits with reservoircomputing. Phys Rev E. 2024;109(2):024210.

 

  1. Akiyama T, Tanaka G. Computational efficiency of multi-steplearning echo state networks for nonlinear time seriesprediction. IEEE Access. 2022;10:28535-28544.

 

  1. Tavakoli SK, Longtin A. Boosting reservoir computerperformance with multiple delays. Phys Rev E.2024;109(5):054203.

 

  1. Bo YC, Wang P, Zhang X. An asynchronously deep reservoircomputing for predicting chaotic time series. Appl SoftComput. 2020;95:106530.

 

  1. Na X, Ren W, Xu X. Hierarchical delay-memory echo statenetwork: a model designed for multi-step chaotic time seriesprediction. Eng Appl Artif Intell. 2021;102:104229.

 

  1. Na X, Zhang M, RenW, Han M. Multistep-ahead chaotic timeseries prediction based on hierarchical echo state networkwith augmented random features. IEEE Trans Cogn Dev Syst.2022;15(2):700-711.

 

  1. Cheng X, Liu Z, Fang H. Deep fusion of discrete-time andcontinuous-time models for long-term prediction of chaoticdynamical systems. Nonlinear Dyn. 2025;113(11):12545-12563.

 

  1. Chepuri R, Amzalag D, Antonsen T, Girvan M. Hybridizingtraditional and next-generation reservoir computing toaccurately and efficiently forecast dynamical systems. Chaos.2024;34(6):063123.

 

  1. Yang L, Pang S, Zhang Y, et al. Improved next generationreservoir computing with time decay factor and kernelfunction. Chaos Solitons Fractals. 2025;198:116514.

 

  1. Lu J, Chen G, Cheng D, Celikovsky S. Bridge the gap betweenthe Lorenz system and the Chen system. Int J Bifurc Chaos.2002;12(12):2917-2926.

 

  1. Qi G, Chen G. Analysis and circuit implementation of a new4d chaotic system. Phys Lett A. 2006;352(4-5):386-397.
Next article in this issue
Share
Back to top
Nonlinear Science and Control Engineering, Published by AccScience Publishing
We use cookies on our website to ensure you get the best experience.
Read more about our cookies here
Accept