ABSTRACT
Differential systems as models of artificial neural networks have been surveyed by Grossberg in [4]. Two kinds of dynamic processes can occur in neural networks: The change of the activation of neurons and the change of the strength of the synaptic connections (weights) between neurons with respect to time. The stability of first kind has been considered previously by Hopfield [6], Hirsch [5], Matsuka [10], and Cohen and Grossberg [2]. The stability of the latter has been investigated by Dong [3]. In each of these investigations, a single Lyapunov function was used to describe the stability requirements.
