Abstract
The classic algorithm of backpropagation of errors requires that the neural transfer function is differentiable and usually the algebraic form of this derivative determines the implementation of the algorithm minimizing the SSE error function. The paper extends the idea of homogeneous ANNs of the feed-forward type, which can be designed with the use of calculus of finite differences. We present a novel model of a neural network which uses a fractional order derivative mechanism. It has been shown that by using numerical approximation of a fractional order derivative, it is possible to smoothly model the dynamics of the transfer function of a single neuron without the need to modify the algebraic form of its base functions like sigmoid. This approach universalizes the neural network model, and enhance the area of possible applications.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Bao, C., Pu, Y., Zhang, Y.: Fractional-order deep backpropagation neural network. Comput. Intell. Neurosci. (2018)
Garrappa, R.: Numerical evaluation of two and three parameter Mittag-Leffler functions. SIAM J. Numer. Anal. 53, 1350–1369 (2015)
Ghosh, U., Sarkar, S., Das, S.: Solution of system of linear fractional differential equations with modified derivative of Jumarie type. Am. J. Math. Anal. 3(3), 72–84 (2015)
Giusti, A.: A comment on some new definitions of fractional derivative. Nonlinear Dyn. 93(3), 1757–1763 (2018). https://doi.org/10.1007/s11071-018-4289-8
Gomolka, Z., Dudek-Dyduch, E., Kondratenko, Y.P.: From homogeneous network to neural nets with fractional derivative mechanism. In: Rutkowski, L., Korytkowski, M., Scherer, R., Tadeusiewicz, R., Zadeh, L.A., Zurada, J.M. (eds.) ICAISC 2017. LNCS (LNAI), vol. 10245, pp. 52–63. Springer, Cham (2017). https://doi.org/10.1007/978-3-319-59063-9_5
Gomolka, Z.: Neurons’transfer function modeling with the use of fractional derivative. In: DepCoS-RELCOMEX 2018, AISC, vol. 761. Springer (2019). https://doi.org/10.1007/978-3-319-91446-6_21
Moret, I.: Shift-and-invert Krylov methods for time-fractional wave equations. Numer. Func. Anal. Optim. 36(1), 86–103 (2015)
Tarasov, V.: No nonlocality. No fractional derivative. Commun. Nonlinear Sci. Numer. Simul. 62, 157–163 (2018)
Wang, J., Wen, Y., Gou, Y., Ye, Z., Chen, H.: Fractional-order gradient descent learning of BP neural networks with Caputo derivative. Neural Networks, 89, 19–30
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2020 Springer Nature Switzerland AG
About this paper
Cite this paper
Gomolka, Z. (2020). Fractional Backpropagation Algorithm – Convergence for the Fluent Shapes of the Neuron Transfer Function. In: Yang, H., Pasupa, K., Leung, A.CS., Kwok, J.T., Chan, J.H., King, I. (eds) Neural Information Processing. ICONIP 2020. Communications in Computer and Information Science, vol 1333. Springer, Cham. https://doi.org/10.1007/978-3-030-63823-8_66
Download citation
DOI: https://doi.org/10.1007/978-3-030-63823-8_66
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-63822-1
Online ISBN: 978-3-030-63823-8
eBook Packages: Computer ScienceComputer Science (R0)