Superior robustness of using power-sigmoid activation functions in Z-type models for time-varying problems solving

Author

Yu-Nong Zhang ; Zhen Li ; Dong-Sheng Guo ; Ke Chen ; Pei Chen

Author_Institution

Sch. of Inf. Sci. & Technol., Sun Yat-sen Univ. (SYSU), Guangzhou, China

Volume

02

fYear

2013

fDate

14-17 July 2013

Firstpage

759

Lastpage

764

Abstract

The performance analyses of Z-type models using PSAF (i.e, power-sigmoid activation functions) for solving the Zhang problems are investigated in this paper. Excellent robustness is demonstrated when using PSAF for very large perturbation errors. Compared with LAF (i.e, linear activation functions), Z-type models using PSAF have better performance on solving not only scalar-valued problems but also matrix-valued (and vector-valued) problems. The two applications finally substantiate the theoretical analysis, and especially show an excellent robustness.

Keywords

matrix algebra; neural nets; perturbation theory; time-varying systems; transfer functions; vectors; PSAF; Z-type models using; linear activation functions; matrix-valued problem; perturbation error; power-sigmoid activation functions; robustness; scalar-valued problem; time-varying problems solving; vector-valued problem; Abstracts; Electronic mail; Vectors; PSAF (power-sigmoid activation functions); Superior robustness; Time-varying problems solving (Zhang problem solving); Very large perturbation errors; Z-type models;

fLanguage

English

Publisher

ieee

Conference_Titel

Machine Learning and Cybernetics (ICMLC), 2013 International Conference on

Conference_Location

Tianjin

Type

conf

DOI

10.1109/ICMLC.2013.6890387

Filename

6890387