COMPLETE CONTROLLABILITY OF CONTINUOUS-TIME RECURRENT NEURAL NETWORKS

Authors
SONTAG E SUSSMANN H
Citation
E. Sontag et H. Sussmann, COMPLETE CONTROLLABILITY OF CONTINUOUS-TIME RECURRENT NEURAL NETWORKS, Systems & control letters, 30(4), 1997, pp. 177-183
Citations number
18
Categorie Soggetti
Controlo Theory & Cybernetics","System Science","Operatione Research & Management Science
Journal title
Systems & control letters → ACNP
ISSN journal
01676911
Volume
30
Issue
4
Year of publication
1997
Pages
177 - 183
Database
ISI
SICI code
0167-6911(1997)30:4<177:CCOCRN>2.0.ZU;2-X
Abstract
This paper studies controllability for the class of control systems co mmonly called (continuous-time) recurrent neural networks. It is shown that, under a generic condition on the input matrix, the system is co ntrollable, for every possible state matrix. The result holds when the activation function is the hyperbolic tangent. (C) 1997 Elsevier Scie nce B.V.