Non-smooth regularization in radial artificial neural networks

9th International Multidisciplinary Scientific and Research Conference "Modern Issues in Science and Technology" Workshop "Advanced Technologies in Aerospace, Mechanical and Automation Engineering"

自然科学;工业技术

Krutikov, V.N.^1 ; Kazakovtsev, L.A.^2 ; Kazakovtsev, V.L.^3

Kemerovo State University, 6 Krasnaya street, Kemerovo

650000, Russia^1

Reshetnev Siberian State University of Science and Technology, 31, Krasnoyarsky Rabochy Av., Krasnoyarsk

660037, Russia^2

Saint Petersburg National Research University of Information Technologies, Mechanics and Optics, 49, Kronverksky Av., Saint Petersburg

197101, Russia^3

关键词: Arbitrary functions;    Initial approximation;    New approaches;    Radial basis functions;    Regularization methods;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/450/4/042010/pdf
DOI  :  10.1088/1757-899X/450/4/042010

来源: IOP

【 摘 要 】

We propose a new approach to the approximation of an arbitrary function by the sum of radial basis functions using non-smooth regularization method combined with a new algorithm in determining the initial approximation which ensures an acceptable quality of approximation uniformly over the data area.