A Novel Neural Network with Transient Chaos and Its Application in Function Optimization
Chunbo Xiu · Computer Engineering and Science · 2006
In this article a novel chaotic neural model whose activation function is composed of Gaussian and Sigmoid functions is proposed. It is shown that the model may exhibit a complex and dynamic property. The most significant bifurcation processes,which lead to chaos, are investigated through the computation of the Lyapunov exponents. Based on this neural model, we propose a novel neural network with transient chaos. It can be applied to solving various complicated optimization problems. Extensive numerical simulations show that the network has a higher ability of searching for globally optimal solutions and has a faster speed.