摘要
介绍了BP神经网络的基本原理,指出了BP算法收敛速度慢、易陷入局部及小值等缺陷,在标准BP算法的基础上引入了几种优化BP算法的方法。针对模式识别应用领域,通过实例,运用Matlab编程对各种较好的网络学习算法的性能进行比较,给出了一个三层BP网络识别含噪声的字母的实例。实验结果表明,改进的BP算法有效地提高了BP算法的收敛速度。
This paper introduces the fundamental of BP neural network, then some defects such as slow convergence rate and easy to get into local minimum in BP algorithm are pointed out, finally, in view of these limitations, several methods are led to optimize BP algorithm. Aiming at pattern recognition application, with Matlab programming tool, this paper studies the training performance of these learning algorithms through comparison. On this basis, this paper offers kinds of programming recognition of using BP network to recognize the noisy English character. Experiment results show that these improved methods increase efficiently the convergence nerformance of BP algorithm.
出处
《科技创业月刊》
2009年第7期148-149,共2页
Journal of Entrepreneurship in Science & Technology
关键词
模式识别
学习算法
BP神经网络
pattern recognition, learning algorithm, BP neural network
