Analysis of chaotic neural networks and system identification

In this thesis, chaotic responses of the chaotic neurons are determined using the various methods. Identification performance of the various neural network models such as feedforward, self feedback, fully recurrent and chaotic neural network is determined for two different nonlinear dynamic systems which are single input-single output and multiple inputs-multiple outputs. Chaotic neural network model needs less input than the feed forward neural model and it is shown that they can be effectively used for system modeling.