摘要
本文在脉冲耦合神经网络 (PCNN PulseCoupledNeuralNetwork)的基础上 ,提出了时延脉冲耦合神经网络(DPCNN DelayPCNN) ,并将其成功地用于求解最短路径 ,同时给出了基于DPCNN的最短路径求解算法 .Caulfield与Kinser提出了用PCNN求解迷宫问题的方法 ,虽然他们的方法也可用于求解最短路径 ,但所需神经元的数量巨大 ,而本文的方法所需的神经元的数量远小于他们的方法 .同时 ,本文的方法充分利用了DPCNN脉冲快速并行传播的特点 ,可迅速地求出最短路径 ,其所需的计算量仅正比于最短路径的长度 ,与路径图的复杂程度及路径图中的通路总数无关 .计算机仿真结果表明 ,采用本文的方法 ,用少量的神经元就可迅速地求出最短路径 .
This paper presents DPCNN (Delay Pulse Coupled Neural Network) based on PCNN and uses DPCNN to find the shortest path successfully. Meanwhile, the algorithm of finding the shortest path based on DPCNN is described. Caulfield and Kinser introduced the PCNN method to solve the maze problem 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 Caulfield and Kinser. In the meantime, due to the pulse parallel transmission characteristic of DPCNN, the approach proposed 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 path graph complexity and the number of existing paths in the graph. The results of computer simulations show that by using the approach proposed in this paper, we can use a small quantity of neurons to find the shortest path quickly.
出处
《电子学报》
EI
CAS
CSCD
北大核心
2004年第9期1441-1443,共3页
Acta Electronica Sinica
基金
国家自然科学基金资助项目 (No 60 1 71 0 36)
国家 863计划基金资助项目 (No 2 0 0 2AA7830 60 )
中国博士后基金资助项目(No 2 0 0 30 342 82 )
关键词
时延PCNN
最短路径
PCNN
Algorithms
Computational complexity
Computer simulation
Optimization
