基于GA-BP神经网络的人脸识别方法研究

Research on the Face Recognition Approach Based on the GA-BP Neural Network

针对传统的BP神经网络算法易陷入局部极小点、训练速度慢的问题,文章用遗传算法(Genetic Algorithm,GA)来优化BP神经网络,实现全局寻优和局部寻优相结合,有效提高神经网络的学习性能和收敛性。实验结果表明,文章提出的方法训练速度快,克服其陷入局部最优的缺点,具有很好的识别性能。

To solve the problem that traditional BP neural network algorithm is easily got stuck in local minima and with low training speed, this paper studies the optimization of BP neural network by using genetic algorithm, which successfully combine global and local optimization and effectively improve the learning ability and convergence of neural network. The experimental results show that the method introduced in this paper has the advantage of fast speed which avoids the network being stuck in local minima as well as good recognition performance.