摘要
BP神经网络具有较强的容错性和自适应学习能力,因而在数字图像识别领域有着广泛的应用。本文在经典BP神经网络的基本算法的基础上,对BP算法的参数设置进行了优化,实现了一种基于分类的改进BP神经网络算法。通过探讨BP神经网络在数字图像分类识别中的应用,详细考察了各种参数对识别效果的影响。实验结果证明改进后的算法有很好的实用价值。
Back-propagation neural network with high fault-tolerance and good adaptive learning ability,is found great applications in digital image recognition.This paper presents an improved algorithm through setting new parameters based on traditional BP algorithm,applies it to digital image recognition and analyzes the influence produced by improved parameters.The experimental result shows that the improved algorithm has a certain practical value.
出处
《计算机与现代化》
2008年第5期17-19,共3页
Computer and Modernization
基金
国家985二期信息创新平台项目(0000-X07204)
关键词
BP神经网络
动量因子
图像识别
BP neural networks
momentum factor
image recognition
