Design of a chaotic feed forward neural network

Several scientific researchers have shown that synapses، far from what considered previously، are one of the most dynamic structures in the nervous system. It varies through and after the training stage. A novel feed forward chaotic neural network is designed in this project. The weights of a trained classic feed forward neural network are considered as basic weights. The outputs of chaotic logistic maps are added to the some of the basic weights and this produces a chaotic feed forward neural network. The chaotic weights are synchronized by a proposed algorithm to reduce the output error. Designed chaotic feed forward neural network can separate some data from test set which contain all the recognition errors of the classic feed forward neural network.