Neural Networks Model for Optimization an Study on Isomorphism Discernment

Abstract

An artificial neural network is composed of large number of simple processing elements by direct links named connections, the benefits of neural networks extend beyond the high computation rates by massive parallelism. Optimization problems could be transferred into a feedback network, the network interconnects the neurons with a feedback path. Graphs isomorphism discernment is one of the most important and difficult issues in graphs theory based structures design. To solve the problem, a Hopfield neural networks (HNN) model is presented in this paper. The solution of HNN is design as a permutation matrix of two graphs, and some operators are improved to prevent premature convergence. It is concluded that the algorithm presented here is efficient for large-scale graphs isomorphism problem and other NP-complete optimization issues.