Regularization of orthogonal neural networks using fractional derivatives

Author

Halawa, Krzysztof

Author_Institution

Wroclaw Univ. of Technol., Wroclaw, Poland

fYear

2009

fDate

20-21 Sept. 2009

Firstpage

74

Lastpage

77

Abstract

A method of regularization of orthogonal neural networks using fractional derivatives is proposed in the paper. The cost function with a penalty for non-smoothness with fractional derivatives enabling to use a priori knowledge. The formula for network weight values which minimize the proposed cost function was derived. It was demonstrated the obtained matrix in normal equations is nonnegative-definite. The results of simulation experiments where the outlined method was used for modeling static nonlinear systems were shown.

Keywords

feedforward neural nets; least squares approximations; a priori knowledge; cost function; fractional derivatives; network weight values; orthogonal neural network regularization; static nonlinear systems modeling; Cost function; Feedforward neural networks; Fourier series; Least squares methods; Neural networks; Neurons; Nonlinear equations; Nonlinear systems; Polynomials; Vectors; Feedforward neural networks; Least squares methods; Modeling; Neural network architecture;

fLanguage

English

Publisher

ieee

Conference_Titel

Information, Computing and Telecommunication, 2009. YC-ICT '09. IEEE Youth Conference on

Conference_Location

Beijing

Print_ISBN

978-1-4244-5074-9

Electronic_ISBN

978-1-4244-5076-3

Type

conf

DOI

10.1109/YCICT.2009.5382425

Filename

5382425