摘要
神经网络的设计主要集中在网络权值的选取和神经网络结构的确定两个方面 ,与遗传算法结合是目前研究发展的趋势 .本文与一般的基于遗传算法的神经网络设计相比 ,提出一个新型算子—— BP算子 ,并对神经网络的权值和结构同时优化 .仿真结果表明该算法结果比较理想 .
Currently research on ANN(Artificial Neutral Network) is mainly focused on the initial weight and structure. It is a developing tendency to combine ANN with GA(Genetic Algorithm). Compared with general ANN design, this paper puts forward a new operator, BP operator, and optimizes the ANN's initial weights and structure at the same time. The result of the emulation is very reasonable.
出处
《系统工程理论与实践》
EI
CSCD
北大核心
2001年第10期42-46,70,共6页
Systems Engineering-Theory & Practice
