A study of functions distribution of neural networks

In this paper, universal learning networks with branch control of relative strength (ULNs with BR) is studied which consists of basic networks and branch control networks. The branch control net work can be used to determine which intermediate nodes of the basic network should be connected to the output node with a coefficient of relative strength ranging from zero to one. This determination will adjust the outputs of the intermediate nodes of the basic network. Therefore, by using ULNs with BC, locally functions distributed networks can be realized depending on the values of the inputs of the network. ULNs with BR is applied to a two-spirals problem. The simulation results show that ULNs with BC exhibits better performance than the conventional networks with comparable complexity.