Abstract
Universal Leaning Network (ULN) which is a super set of supervised learning networks has been already proposed. Parameters of ULNs are trained in order to optimize a criterion function as conventional neural networks, and after training they are fixed as constant parameters. In this paper a new method to alter the parameters, therefore in a special case, to control the branch connection depending on the network flows is presented to enhance flexibility of the networks. In the proposed method, there exist two kinds of networks, the first one is a basic network which includes varying parameters and the other one is a network which cal-culates the optimal varying parameters, therefore decides the branch connection in a special case depending on the network flows of the basic network.
From simulations where parameters of a neural network are altered and branch connection in the neu-ral network is determined by a fuzzy inference network, it is shown that the proposed network has higher representation abilities than the conventional networks.
