Abstract
We consider the solution of boundary value problems of mathematical physics with neural networks of a special form, namely radial basis function networks. This approach does not require one to construct a difference grid and allows to obtain an approximate analytic solution at an arbitrary point of the solution domain. We analyze learning algorithms for such networks. We propose an algorithm for learning neural networks based on the method of trust region. The algorithm allows to significantly reduce the learning time of the network.
Similar content being viewed by others
References
Tolstykh, A.I. and Shirobokov, D.A., Mesh-Free Method Based on Radial Basis Functions, Zh. Vychisl. Mat. Mat. Fiz., 2005, vol. 45, no. 8, pp. 1498–1505.
Liu, G.R., Mesh-Free Methods: Moving Beyond the Finite Element Method, Boca Raton: CRC Press, 2003.
Kansa, E.J., Motivation for Using Radial Basis Function to Solve PDEs. http://www.cityu.edu.hk/rbf-pde/files/overview-pdf.pdf
Buhmann, M.D., Radial Basis Functions: Theory and Implementations, Cambridge: Cambridge Univ. Press, 2004.
Fasshauer, G.E., Meshfree Approximation Methods with MATLAB, River Edge: World Scientific, 2007.
Chen, W. and Fu, Z.J., Recent Advances in Radial Basis Function Collocation Methods, New York: Springer, 2014.
Fornberg, B. and Flyer, N., Solving PDEs with Radial Basis Functions, Acta Numerica, 2015, vol. 4, pp. 215–258.
Jianyu, L., Siwei, L., Yingjian, Q., and Yaping, H., Numerical Solution of Differential Equations by Radial Basis Function Neural Networks, Proc. Int. Joint Conf. on Neural Networks, 2002, vol. 1, pp. 773–777.
Mai-Duy, N. and Tran-Cong, T., Solving High Order Ordinary Differential Equations with Radial Basis Function Networks, Int. J. Numer. Methods Eng., 2005, vol. 62, pp. 824–852.
Sarra, S., Adaptive Radial Basis Function Methods for Time Dependent Partial Differential Equations, Appl. Numer. Math., 2005, vol. 54, no. 1, pp. 79–94.
Chen, H., Kong, L., and Leng, W., Numerical Solution of PDEs via Integrated Radial Basis Function Networks with Adaptive Training Algorithm, Appl. Soft Comput., 2011, vol. 11, no. 1, pp. 855–860.
Kumar, M. and Yadav, N., Multilayer Perceptrons and Radial Basis Function Neural Network Methods for the Solution of Differential Equations: A Survey, Comput. Math. Appl., 2011, vol. 62, pp. 3796–3811.
Yadav, N., Yadav, A., and Kumar, M., An Introduction to Neural Network Methods for Differential Equations, New York: Springer, 2015.
Tarkhov, D.A., Neirosetevye modeli i algoritmy. Spravochnik (Neural Network Models and Algorithms. Reference), Moscow: Radiotekhnika, 2014.
Gorbachenko, V.I. and Artyukhina, E.V., Mesh-Free Methods and Their Implementation with Radial Basis Neural Networks, Neirokomp’yut.: Razrabotka, Primen., 2010, no. 11, pp. 4–10.
Gorbachenko, V.I. and Zhukov, M.V., Solving Boundary Value Problems of Mathematical Physics Using Radial Basis Function Networks, Comput. Math. Math. Phys., 2017, vol. 57, no. 1, pp. 145–155.
Goodfellow, I., Bengio, Y., and Courville, A., Deep Learning, Boston: MIT Press, 2017. Translated under the title Glubokoe obuchenie, Moscow: DMK Press, 2018.
Jia, W., Zhao, D., Shen, T., Su, C., Hu, C., and Zhao, Y., A New Optimized GA-RBF Neural Network Algorithm, Comput. Intelligence Neurosci., 2014, Article ID982045.
Gill, P.E., Murray, W., and Wright, M.H., Practical Optimization, London: Academic, 1981. Translated under the title Prakticheskaya optimizatsiya, Moscow: Mir, 1985.
Sutskever, I., Martens, J., Dahl, G., and Hinton, G., On the Importance of Initialization and Momentum in Deep Learning, Proc. 30th Int. Conf. on Machine Learning, 2013, vol. 28, pp. 1139–1147.
Alkezuini, M.M. and Gorbachenko, V.I., Training Networks of Radial Basis Functions with Nesterov’s Method for Solving Boundary Problems of Mathematical Physics, Proc. XII Intl. Sci.-Tech. Conf. Analytic and Numerical Methods of Modeling Natural Science and Social Problems, Penza: PGU, 2017, pp. 171–175.
Fletcher, R. and Reeves, C.M., Function Minimization by Conjugate Gradients, Comput. J., 1964, vol. 7, pp. 149–154.
Polak, E. and Ribiére, G., Note sur la convergence de méthodes de directions conjugués, Revue Française d’Inform. Recherche Opération., Série Rouge, 1969, vol. 3, no. 1, pp. 35–43.
Zhang, L., Li, K., He, H., and Irwin, G.W., A New Discrete-Continuous Algorithm for Radial Basis Function Networks Construction, IEEE Trans. Neural Networks Learning Syst., 2013, vol. 24, no. 11, pp. 1785–1798.
Xie, T., Yu, H., Hewlett, J., Rozycki, P., and Wilamowski, B., Fast and Efficient Second-Order Method for Training Radial Basis Function Networks, IEEE Trans. Neural Networks Learning Syst., 2012, vol. 23, no. 4, pp. 609–619.
Markopoulos, A.P., Georgiopoulos, S., and Manolakos, D.E., On the Use of Back Propagation and Radial Basis Function Neural Networks in Surface Roughness Prediction, J. Indust. Engin. Int., 2016, vol. 12, no. 3, pp. 389–400.
Zhang, L., Li, K., and Wang, W., An Improved Conjugate Gradient Algorithm for Radial Basis Function (RBF) Networks Modelling, in Proc. UKACC Int. Conf. on Control, 2012, pp. 19–23.
Sadeghi, M., Pashaie, M., and Jafarian, A., RBF Neural Networks Based on BFGS Optimization Method for Solving Integral Equations, Adv. Appl. Math. Biosci., 2016, vol. 7, no. 1, pp. 1–22.
Conn, A.R., Gould, N.I.M., and Toint, P.L., Trust-Region Methods, MPS-SIAM, 1987.
Wild, S.M., Regis, R.G., and Shoemaker, C.A., ORBIT: Optimization by Radial Basis Function Interpolation in Trust-Regions, SIAM J. Scientific Comput., 2008, vol. 30, no. 6, pp. 3197–3219.
Bernal, F., Trust-Region Methods for Nonlinear Elliptic Equations with Radial Basis Functions, Comput. Math. Appl., 2016, vol. 72, no. 7, pp. 1743–1763.
Gorbachenko, V.I. and Zhukov, M.V., The Learning of Radial Basis Function Network Using Trust Region Method to Solve Poisson’s Equation, Inform. Tekhnol., 2013, no. 9, pp. 65–70.
Brink, H., Richards, J.W., and Fetherolf, M., Real-World Machine Learning, Shelter Island: Manning, 2016. Translated under the title Mashinnoe obuchenie, St. Petersburg: Piter, 2017.
Steihaug, T., The Conjugate Gradient Method and Trust Region in Large Scale Optimization, SIAM J. Numer. Anal., 1983, vol. 20, no. 3, pp. 626–637.
Watkins, D.S., Fundamentals of Matrix Computations, New York: Springer, 1991. Translated under the title Osnovy matrichnykh vychislenii, Moscow: Binom, 2012.
Author information
Authors and Affiliations
Corresponding author
Additional information
Original Russian Text © L.N. Elisov, V.I. Gorbachenko, M.V. Zhukov, 2018, published in Avtomatika i Telemekhanika, 2018, No. 9, pp. 95–105.
Rights and permissions
About this article
Cite this article
Elisov, L.N., Gorbachenko, V.I. & Zhukov, M.V. Learning Radial Basis Function Networks with the Trust Region Method for Boundary Problems. Autom Remote Control 79, 1621–1629 (2018). https://doi.org/10.1134/S0005117918090072
Received:
Published:
Issue Date:
DOI: https://doi.org/10.1134/S0005117918090072