Title :
Some stability properties of recurrent neural networks
Author :
Si, Jennie ; Lin, Ching-Fang
Author_Institution :
Dept. of Electr. Eng., Arizona State Univ., Tempe, AZ, USA
fDate :
29 June-1 July 1994
Abstract :
Stability properties of recurrent neural networks are investigated using Lyapunov stability theory and functional analytic means. Sufficient conditions for the global asymptotic stability and exponentially asymptotic stability of equilibrium points of a class of recurrent neural networks are provided. The results obtained can be applied when recurrent neural networks are used as computation models, in particular as optimization models. The results may also be used as stability analysis tools for some class of nonlinear control systems.
Keywords :
Lyapunov methods; asymptotic stability; functional analysis; recurrent neural nets; Lyapunov stability; equilibrium points; exponentially asymptotic stability; functional analytic means; global asymptotic stability; nonlinear control systems; optimization models; recurrent neural networks; sufficient conditions; Artificial neural networks; Asymptotic stability; Computer networks; Lyapunov method; Nonlinear control systems; Nonlinear dynamical systems; Power system modeling; Recurrent neural networks; Stability analysis; Sufficient conditions;
Conference_Titel :
American Control Conference, 1994
Print_ISBN :
0-7803-1783-1
DOI :
10.1109/ACC.1994.735194
