基于改进BP神经网络的样本分类方法

Research and Application of Improved BP Neural Network Algorithm

为了应对BP神经网络在复杂样本分类过程中存在的因为网络结构复杂,网络跳转多样而导致的分类能力低,分类结果模糊等问题,提出了基于因子分析和遗传算法的BP神经网络。本算法首先通过因子分析的方法降低BP神经网输入样本的维度,然后使用遗传算法改进BP神经网络的分析过程,从而实现BP神经网络的强分类分析能力。

To handle the problem of low classification ability and diverse results in complex sample classification process because of the complex network structure. The factor analysis and genetic algorithm based BP neural network is proposed. Firstly,dimension of the input samples of BP neural network are reduced by factor analysis algorithm. Then genetic algorithm is used to improve the analysis process of BP neural network so as to realize the strong classification analysis ability of BP neural network.