Application of Genetic Algorithm to Optimization of BP Neural Network

摘要：

To alleviate the shortcoming of easily sinking into the local minimum existing in the BP neural network, the paper exploits the genetic algorithm to optimize the BP neural network. First of all, the genetic algorithm is utilized to optimize the weight values as well as the threshold values of the BP neural network. Subsequently, by using the optimized weight values and threshold values, we are able to get the improved BP neural network. Furthermore, we employ the simulation data to measure the performance of the improved BP neural network. The numerical results indicate that the optimized BP neural network can effectively overcome the local minimum of the original BP neural network and outperform the original BP neural network in the aspects of convergence speed and computation accuracy.

关键词： BP neural network genetic algorithm optimization

作者: Liming Xie

作者单位: School of Information and Control Engineering, Weifang University, Weifang, Shandong, China

会议类型: 国际会议

会议名称: 2010 International Conference on Circuit and Signal Processing(2010年电路与信号处理国际会议 ICCSP 2010)

会议地点: 上海

会议语种:英文

页码: 579-581

在线出版日期: 2010-12-25（万方平台首次上网日期，不代表论文的发表时间）