摘要
模型将GA、SA与BP 3种算法有机地融合在一起,实现优势互补。采用二进制与实数混合编码,可以动态地根据样本特征对BP网络中的输入节点数、隐层节点数、转移函数、权值与阈值等进行自适应优化调整。在保证精度的前提下,采用较少的输入节点和隐层节点数,使网络的结构相对简单。采用自适应交叉率、变异率与学习率,以增强网络的自适应与泛化能力,极大地减少人为主观因素对网络设计的影响。
In this paper,a dynamic all parameters adaptive BP neural networks model is proposed by integration of genetic algorithms,simulated annealing and error back propagation to achieve supplementary advantages.Using binary mixture with real code,the model can dynamic and adaptive optimize the input nodes,hidden nodes, transfer function,weights and bias of BP networks according to samples characteristic.Under the premise of ensuring accuracy,the architecture of the network model is relatively simple(less input and hidden nodes) to improve the adaptation and generalization ability of networks,and to greatly reduce the impact of human factors on networks design.
出处
《系统管理学报》
北大核心
2008年第5期499-503,共5页
Journal of Systems & Management
基金
国家自然科学基金资助项目(70573101)
关键词
动态
全参数
自适应
遗传算法
BP网络
结构确定
dynamic
all parameters
adaptive
genetic algorithms
BP neural network
structure(determination)