基于遗传BP算法的手写体识别

神经网络具有模拟人类的大脑活动、良好的自学习、自适应、联想记忆、并行处理和非线形转换的能力.本文阐述了BP神经网络基本原理以及BP网络手写体识别模型,研究分析了BP神经网络手写体识别模型的缺陷并提出了优化策略.在此基础上,提出一种基于改进结构的BP神经网络来实现手写体数字识别方案,除了改进BP网的结构外,还对网络学习算法进行了改进,采用了BP和GA相结合的算法,提高了网络的学习训练速度和识别效果.

The Handwritten Characters Recognition based on GA-BP Algorithm

This paper discusses the principle of artificial neural networks, BP algorithms and the recognition method of handwritten characters, base on neural networks. To improve the training efficiency and identifying rate, this method improves the net structure and learning algorithms, using the genetic algorithms and BP algorithms.