The Bifurcation of a Class of Discrete-Time Neural Networks with a General Activation Function
Weina Chen · Chinese Annals of Mathematics · 2009
By using classical bifurcation theories,the authors investigate a class of discretetime neural networks with a general activation function,and obtain the sufficient condition of period-doubling bifucation and saddle-node bifucation of this model,which can be regarded as an extension of a sinusoidal activation function.As a result,an important theoretical foundation for the application of this class of neural networks is provided.