关键词: 神经网络 BP算法 BP改进 网络粒子群算法

Abstract:

A BP neural network learning algorithm based on improved Particle Swarm Optimization（PSO） is proposed in this paper. In this algorithm, the conventional BP algorithms are improved first;and the crunodes of the input layer, the hidden layer and the output layer in the neural network reach an optimum solution. However, the gradient descent method in traditional BP algorithms is substituted by the PSO algorithm,and the improved algorithm does  ly trap into a local minima and has a better generalization performance. The algorithm is applied to the application design of the stock prediction for   the Shanghai Stock indexes. The results show that this algorithm can reduce the number of iterations obviously,and increase the precision of convergence   , and its generalization performance is superior to the traditional BP algorithms.

Key words: neural network, BP algorithm, BP improvement, PSO algorithm