摘要
设计并实现了一种基于改进BP神经网络的字符识别系统,可克服传统BP算法最终进化至最优解较慢和神经网络易陷入局部最小值的缺陷,具有较好的全局性和收敛速度,经过实验,具有较好的识别速度和精度。
This article designed and realized one kind of the character recognition system which based on improving the BP neural network, could overcome the problem that the traditional BP algorithm evolved to the optimal solution slowly and the flaw that the neural network failed into the local minimum easily. The improved BP neural network had the good overall importance and the convergence rate. The process tested this system having the good recognition speed and the precision.
出处
《装备制造技术》
2009年第2期90-92,共3页
Equipment Manufacturing Technology
基金
太原市科技兴市专项基金资助项目(08122071)
关键词
神经网络
字符识别
BP算法
neural network
character recognition
back propagation algorithm
