摘要
基于进化算法可有效解决神经网络设计和实现中存在的一些问题 ,使网络具有更优的性能。在此对基于进化计算的神经网络设计和实现的研究内容及进展情况进行综述 ,讨论了网络实现的关键问题 ,包括网络权重的进化训练 ,网络结构进化设计 ,学习规则进化选取以及进化操作算子设计等 ,并分析了相关的研究和发展方向。
Design and implementation of artificial neural networks based on evolutionary computation can lead to significantly better network performance. A review on evolutionary computation based neural networks design and implementation is presented in details and some related aspects are discussed emphatically. Among them are the evolving training of network weights, the evolving design of network architecture, the evolving design of learning rule, different search operators, and the possible future research directions in this area.
出处
《控制与决策》
EI
CSCD
北大核心
2001年第3期257-262,共6页
Control and Decision
