摘要
神经网络能精确地对复杂问题进行预测,但易受训练过度的影响且训练速度慢。遗传算法是一种全局优化搜索算法,这种方法的要求是拟合度函数必须收敛于最小误差,这为具体的实现方案留下了许多自由空间。将遗传算法与神经网络结合起来使用可以在较高的层次上提高模型的可理解性。
The nerve network can by the square carry on an estimate to the complicated problem,but easily receive training to do excesssive influence and train speed slow.Inheritting the calculate way is that a kind of overall situation is excellent to turn to search calculate way,the request of this method is to draw up to match one degree function have to be refrained from rash action to a minimum error margin,this for concretely carried out a project to leave many free spaces.Put together heredity calculate way and nerve network knot use can raise the apprehensible of the model on the higher layer.
出处
《廊坊师范学院学报(自然科学版)》
2011年第2期40-43,共4页
Journal of Langfang Normal University(Natural Science Edition)
关键词
遗传算法
计算机辅助设计
人工神经网络
流体机械
inherit calculate way
the calculator lends support to a design
artificial nerve network
fluid machine