Delay PCNN and its application for optimization

Abstract

This paper introduces the DPCNN (Delay Pulse Coupled Neural Network) based on the PCNN and uses the DPCNN to find the shortest path. Cauflield and Kinser introduced the PCNN method to solve the maze[1] and although their method also can be used to find the shortest path, a large quantity of neurons are needed. However, the approach proposed in this paper needed very fewer neurons than proposed by Cauflield and Kinser. Meanwhile, due to the parallel pulse transmission characteristic of the DPCNN, our approach can find the shortest path quickly. The computational complexity of our approach is only related to the length of the shortest path, and independent to the weighted graph complexity and the number of existed paths in the graph. © Springer-Verlag 2004.