G-S chaotic neural networks and its applications
Guowei Xu · 2010
A novel G-S chaotic neural networks is proposed based on G-S chaotic neuron, whose structure is similar to BP neural networks. The activation function is no monotonous function and the neurons have two states. In the process of learning, the states of neurons are chaotic. According to the states, the weights can be adjusted. In the process of working, the sates of neurons are not chaotic. The learning algorithm of the networks is a chaos optimization method, which can get over the disadvantages of conventional networks. The function approximation and the hysteresis modeling of piezoelectric can be resolved by the networks. The experiment results proved the validity of the algorithm.