The Characteristic Analysis And The Design Of Discrete Hopfield Neural Networks

In this paper, the method of directly convergent equation, a method on how to find the values of the network parameters, is put forward from the inner running mechanism of Neural Networks. Compared with the traditional methods, the Hebb regulation is not taken into account in this method ,nor is the energy function involved. Because the convergent equations are inequality groups, there are many groups of the values of the networks parameters that can meet the equations. And with the different group, the Neural Networks have different characteristics. Therefore, it is very necessary to make clear the relations between the values of the network parameters and the characteristics of the Neural Networks.The method of analysis is used in this paper. Fistly, the relations between the values of network parameters and the characteristics of single neuron are studied. Many results about the characteristics of single neuron are gained. And the primary regulations on how to find the values of network parameters are given. Then, the relations between network parameters and the characteristics of the Neural Networks are studied. Characteristic equations of three kinds of Neural Networks are gained. Through the analysis to the substance of running equations of networks, the evolutive theorem of Neural Networks was proved.