摘要
遗传算法在复杂问题的优化搜索方面具有很好的健壮性和性能。文中模拟自然界的进化过程,试图使用遗传算法搜索适当的解决问题的神经元网络结构。通过构造一个自进化系统,从单个神经元个体构成的初始群体出发,模拟自然界中个体的学习、变异、繁殖,演示系统中群体的自进化过程,找出解决逻辑运算问题的适当的神经元网络结构。实验结果表明,与标准的BP网络比较,使用遗传算法的自进化系统能够找到解决逻辑运算问题更优的网络结构。
Genetic algorithm can effectively solve many complex problems through optimized searching.This paper tried to search the proper neural network structure which could solve the problem by simulating the evolution process of nature using genetic algorithm.Through building a self-evolution system,the proper structure of the neural network that can accomplish logical computations may be found by beginning with a single neuron and then self-evolution.The experimental results indicate that genetic algorithm can help finding a better structure of neural networks than only standard BP network.
出处
《计算机技术与发展》
2010年第9期86-89,共4页
Computer Technology and Development
基金
高效能服务器和存储技术国家重点实验室开放基金项目(2009HSSA08)
关键词
遗传算法
自进化
网络结构
genetic algorithm
self-evolution
network structure
