Abstract
This paper presents a novel machine learning method for predicting chaotic spatiotemporal systems by proposing a parallel reservoir-like neural network. In contrast to previous machine learning methods, which require big data and suffer from high computing costs, the main advantage of the method is that it is based on short-term data and only needs to train the weight of the output layer. Theoretically, the ratio of training data length and the prediction data length can reach 2:1. First, the method for transforming an infinite-dimensional system into a finite-dimensional system is introduced. Then, based on this concept and the spatiotemporal information transformation, a network training algorithm of spatial parallel autoreservoir computing structure is proposed. Numerical experiments verified that with short-term data, the proposed method performs better than some widely used data-driven methods.
Similar content being viewed by others
Code Availability Statement
The code used in this study is available at https://github.com/413582273/SPARNN.
References
Bengio Y, Simard P, Frasconi P (1994) Learning long-term dependencies with gradient descent is difficult. IEEE Trans Neural Netw 5(2):157–166. https://doi.org/10.1109/72.279181
Brenner M, Eldredge J, Freund J (2019) Perspective on machine learning for advancing fluid mechanics. Phys Rev Fluids. https://doi.org/10.1103/PhysRevFluids.4.100501
Brunton SL, Noack B, Koumoutsakos P (2020) Machine learning for fluid mechanics. In: Davis S, Moin P(eds) Annual review of fluid mechanics, 52: 477–508. https://doi.org/10.1146/annurev-fluid-010719-060214
Carroll TL (2018) Using reservoir computers to distinguish chaotic signals. Phys Rev E. https://doi.org/10.1103/PhysRevE.98.052209
Chen C, Liu Z, Feng S (2018) Universal approximation capability of broad learning system and its structural variations. IEEE Trans Neural Netw Learn Syst 30(4):1191–1204. https://doi.org/10.1109/TNNLS.2018.2866622
Chen P, Liu R, Aihara K, Chen L (2020) Autoreservoir computing for multistep ahead prediction based on the spatiotemporal information transformation. Nat Commun. https://doi.org/10.1038/s41467-020-18381-0
Fan H, Jiang J, Zhang C, Wang X, Lai Y (2020) Long-term prediction of chaotic systems with recurrent neural networks
Kawakami K (2008) Supervised sequence labelling with recurrent neural networks. Ph. D. thesis
Haynes ND, Soriano MC, Rosin DP, Fischer I, Gauthier DJ (2015) Reservoir computing with a single time-delay autonomous Boolean node. Phys Rev E. https://doi.org/10.1103/PhysRevE.91.020801
Hochreiter S, Schmidhuber J (1997) Long short-term memory. Neural Comput 9(8):1735–1780. https://doi.org/10.1162/neco.1997.9.8.1735
Jiang J, Lai YC (2019) Model-free prediction of spatiotemporal dynamical systems with recurrent neural networks: Role of network spectral radius. Phys Rev Res. https://doi.org/10.1103/PhysRevResearch.1.033056
Larger L, Baylon Fuentes A, Martinenghi R, Udaltsov VS, Chembo YK, Jacquot M (2017) High-speed photonic reservoir computing using a time-delay-based architecture: million words per second classification. Phys Rev X. https://doi.org/10.1103/PhysRevX.7.011015
Lu Z, Pathak J, Hunt B, Girvan M, Brockett R (2017) Reservoir observers: model-free inference of unmeasured variables in chaotic systems. Chaos 10(1063/1):4979665
Pathak J, Lu Z, Hunt BR, Girvan M, Ott E (2017) Using machine learning to replicate chaotic attractors and calculate Lyapunov exponents from data. Chaos 10(1063/1):5010300
Pathak J, Wikner A, Fussell R, Chandra S, Hunt BR, Girvan M, Ott E (2018) Hybrid forecasting of chaotic processes: using machine learning in conjunction with a knowledge-based model. Chaos 10(1063/1):5028373
Sirignano J, Spiliopoulos K (2018) Journal of computational physics. Ann Rev Fluid Mech. https://doi.org/10.1016/j.jcp.2018.08.029
Takens F (1981) Detecting strange attractors in turbulence. In: David R, Sang YL (eds) Dynamical systems and turbulence. Springer, Berlin, pp 366–381
Yuan X, Li L, Wang Y (2020) Nonlinear dynamic soft sensor modeling with supervised long short-term memory network. IEEE Trans Industr Inf 16(5):3168–3176. https://doi.org/10.1109/TII.2019.2902129
Acknowledgements
This research is supported by the National Natural Science Foundation of China (Nos. U1806203 and 61533011).
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors declare that they have no conflict of interest.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Wang, Y., Liu, S. Short-term data-based spatial parallel autoreservoir computing on spatiotemporally chaotic system prediction. Neural Comput & Applic 34, 8713–8722 (2022). https://doi.org/10.1007/s00521-021-06854-2
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00521-021-06854-2