Abstract
The paper aims to present a description of the tm-flow library for the “flowly” construction and training of polynomial neural networks (PNN) for time-evolving process prediction. The introduced polynomial models have a strong relationship to dynamic systems that can be described by a system of nonlinear differential equations. The paper provides implementation details and an explanation of training strategies, along with a few illustrative numerical examples. The source code is available at https://github.com/PNN-Lab/tmflow/.
Supported by Saint Petersburg State University, project ID: 94029367.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Abadi, M., Agarwal, A., Barham, P., Brevdo, E.: TensorFlow: large-scale machine learning on heterogeneous systems (2015). https://www.tensorflow.org/
Al-Jumeily, D., Ghazali, R., Hussain, A.: Predicting physical time series using dynamic ridge polynomial neural networks. PLoS ONE 9(8), e105766 (2014)
Chollet, F.: Deep Learning with Python. Manning Publications Co., New York (2018)
Dubey, A., Radenovic, F., Mahajan, D.: Scalable interpretability via polynomials, June 2022
Golovkina, A., Kozynchenko, V.: Neural network representation for ordinary differential equations. In: Dolinina, O., et al. (eds.) AIES 2022. SSDC, vol. 457, pp. 39–55. Springer, Cham (2023). https://doi.org/10.1007/978-3-031-22938-1_3
Huang, C.T., Chen, J.C., Wu, J.L.: Learning sparse neural networks through mixture-distributed regularization. In: 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), Seattle, WA, USA, pp. 2968–2977. IEEE, June 2020
Ivakhnenko, A.G.: Polynomial theory of complex systems. Trans. Syst. Man Cybern. 4, 364–378 (1971)
Klimenko, I.S.: Implementation of the matrix mapping method for solving a system of differential equations. In: Control Processes and Stability, Saint Petersburg, Russia, vol. 9, pp. 53–57 (2022)
Liu, J., Xu, Z., Shi, R., Cheung, R.C.C., So, H.K.H.: Dynamic sparse training: find efficient sparse network from scratch with trainable masked layers, April 2020
Louizos, C., Welling, M., Kingma, D.P.: Learning sparse neural networks through \$L_0\$ regularization, June 2018
Oh, S.K., Pedrycz, W., Park, B.J.: Polynomial neural networks architecture: analysis and design. Comput. Electr. Eng. 29(6), 703–725 (2003)
Shin, Y., Ghosh, J.: The pi-sigma network: an efficient higher-order neural network for pattern classification and function approximation. In: IJCNN-1991-Seattle International Joint Conference on Neural Networks, vol. 1, pp. 13–18 (1991). https://doi.org/10.1109/IJCNN.1991.155142
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2023 The Author(s), under exclusive license to Springer Nature Switzerland AG
About this paper
Cite this paper
Klimenko, I., Golovkina, A., Ruzhnikov, V. (2023). Polynomial Neural Layers for Numerical Modeling of Dynamical Processes. In: Gervasi, O., et al. Computational Science and Its Applications – ICCSA 2023 Workshops. ICCSA 2023. Lecture Notes in Computer Science, vol 14109. Springer, Cham. https://doi.org/10.1007/978-3-031-37120-2_17
Download citation
DOI: https://doi.org/10.1007/978-3-031-37120-2_17
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-031-37119-6
Online ISBN: 978-3-031-37120-2
eBook Packages: Computer ScienceComputer Science (R0)