Non-Linear Mapping Approaching Research And Application Based On The Two Weights Neural Networks

Two weights neurons are a type of higher neurons with different characteristics in each direction and a good property of local reaction. The feedback-forward neural network made up of it is a functional approaching network of advantageous properties. However, since the model of two weights neuron came into being, there has been less research on functional approaching. This paper, aiming at the problem of forecasting the time series of chaos that usually run into in the actual field of engineering application, studies the theory of nonlinear reflection approaching of two weights neural networks and its ability to forecast the time series of chaos forecast.This paper, based on comprehensive sum-up of the existing research findings in functional approaching of feedback-forward neural network, constructs a new structure of neural network----a structure of two weights neural network based on operators, does some research on the theory of nonlinear reflection approaching, proves that it can achieve the approaching to a continuous nonlinear feedback system to any degree and...