Linear Threshold Discrete-Time Recurrent Neural Networks: Stability and Globally Attractive Sets
Tao Shen, Ian R. Petersen · IEEE Transactions on Automatic Control · 2015
The stability of linear threshold dynamic neural networks is studied, and a series of methods to obtain globally attractive sets is proposed. A sufficient condition to judge whether an invariant set is a globally attractive set is also proposed. This method requires only the solution to a class of linear matrix inequalities. Also, two direct methods to obtain globally attractive sets are given. The stability criteria presented are based on the proposed globally attractive sets. Some numerical examples are given to illustrate the effectiveness of the obtained results.