基于AHGA+BP的神经网络混合学习策略

在自适应递阶遗传算法(AHGA)中融入BP操作可以在对神经网络结构进行优化的同时,充分利用遗传算法的全局寻优能力在大范围内搜索可能的误差极值区域,并利用BP算法沿误差最速下降方向在极值点附近快速搜索,从而达到全局最优和快速搜索的有机结合。通过对混沌时序信号的预测,表明了混合学习策略在较大程度上改进了神经网络的学习性能和泛化能力。

The Hybrid Learning Strategy of Neural Network Based on AHGA+BP Algorithm

BP operation is introduced into adaptive hierarchical genetic algorithm.So it makes full use of the global optimization of GA to search the possible extreme area in a large scope and it uses BP to fast search near the extreme point.At the same time,the structure of network can be optimized because of hierarchical coding.Then global optimization and fast search are combined.The hybrid learning strategy is applied in the subject of"the chaotic time series prediction".It demonstrates that hybrid learning strategy improves mostly the learning performance and generalization ability of neural network.