| 基因遗传算法用于人工神经网络的训练 |
| |
| 引用本文： | 陈吉红,师汉民,陈日曜.基因遗传算法用于人工神经网络的训练[J].华中科技大学学报(自然科学版),1992(Z1). |
| |
| 作者姓名： | 陈吉红 师汉民 陈日曜 |
| |
| 作者单位： | 华中理工大学机械工程一系
(陈吉红，师汉民)，华中理工大学机械工程一系(陈日曜) |
| |
| 摘 要： | 根据基因遗传算法GA的基本原理,阐述了GA用于神经网络训练问题的适合度函数和编码方法,分析了标准GA的不足,提出了自适应浮动区间中心点、全局搜索和局部精确搜索等改进策略,发展了GENITOR算法.训练实例表明,GA不仅能代替BP算法,而且能完成神经元激发函数不可微的多层神经网络的训练.
|
| 关 键 词： | 基因遗传算法 GENITOR算法 神经网络 BP算法 切屑类型 |
Application of the Genetic Algorithm in Neural Network Training |
| |
| Chen Jihong Shi Hanmin Chen Riyao.Application of the Genetic Algorithm in Neural Network Training[J].JOURNAL OF HUAZHONG UNIVERSITY OF SCIENCE AND TECHNOLOGY.NATURE SCIENCE,1992(Z1). |
| |
| Authors: | Chen Jihong Shi Hanmin Chen Riyao |
| |
| Institution: | Chen Jihong Shi Hanmin Chen Riyao |
| |
| Abstract: | The principle of the genetic algorithm is described. The fitness and coding method of the genetic algorithm for neural network training are discussed. Some new search strategies including the adaptive center point, global search and local accurate search are proposed to improve the GENITOR algorithm. Experimental results show that the genetic algorithm can not only replace the BP algorithm for neural network training,but can also be used to train those neural networks with the neuron activation function undifferentiable. |
| |
| Keywords: | neural network back propagation algorithm genetic algorithm GENITOR algorithm chip type |
| 本文献已被 CNKI 等数据库收录！ |
