用于英文字母识别的三种人工神经网络的设计  

Design of three artificial neural networks used on English character recognition

作　　者：丛爽[1] 陆婷婷[1] 

机构地区：[1]中国科学技术大学自动化系,合肥230027

出　　处：《仪器仪表学报》2006年第z3期2242-2244,共3页Chinese Journal of Scientific Instrument

基　　金：国家自然科学基金资助项目(50375148)

摘　　要：分别采用单层感知器、BP网络和霍普菲尔德网络三种神经网络对26个英文字母进行识别,通过实验给出各网络的识别出错率,从中可以看出三种人工神经网络各自的优缺点,为选择神经网络进行字符识别提供了一定的依据。The perception, BP Networks and Hopfield Networks are designed to recognize the 26 English letters. The recognition error rates are given by the result of experiment, from which one can see the advantages and disadvantages of these three neural networks. And the study and analysis in this paper are useful for selecting ANN to recognize character.

关 键 词：性能对比 感知器 BP网络 霍普菲尔德网络 字符识别 

分 类 号：TH7-55[机械工程—仪器科学与技术]