A Discrete-Time Recurrent Neural Network for Solving Shortest Path problem

摘要

This paper presetns a discrete-time recurrent neural network for solving the shortest path problem. The proposed discrete-time recurrent neural network is proven to be globally convergent to an exact solution. In addition, the proposed neural network has fixed design parameters adn simple architecture, thus is more suitable for hardware implementation. Furthermore, an improved network with a larger step size is proposed to increase the convergence rate. The performane and operating characteristics of the proposed neural network are demonstrated by means of simulation results.