摘要

In this paper, the dynamic behaviours of the asymmetric Cohen-Grossberg neural networks are studied, and some sufficient conditions for the local and global stability of the networks are proposed. Based on the stability results and recently developed system designing method, the networks are constructed for storing and retrieving binary and non-binary patterns, and the network performances are analysed by numerical simulations. It is shown that the designed networks can act as information retrieval systems.

全文