基于BP神经网络的手写字符识别  

Handwritten Character Recognition Based on BP Neural Network

作　　者：陆文玲 LU Wen-ling(Nanjing Forestry University,Nanjing Jiangsu 210037)

机构地区：[1]南京林业大学,江苏南京210037

出　　处：《数字技术与应用》2021年第1期64-66,共3页Digital Technology & Application

摘　　要：文章对比了使用TensorFlow搭建的两种使用MNIST数据集训练的神经网络模型,并对神经元个数、激活函数、学习率、优化器等参数不断优化。结果表明,在足够的训练数据集下,简单的神经网络模型对手写字符的识别已经能达到较高水平,额外的技巧则大大增强了模型的泛化能力与鲁棒性。The article compares twoneural network models trained using TensorFlow on MNIST datasets,and optimizes parameters such as the number of neurons,activation function,learning rate,and optimizer.The results show that the simple neural network model can achieve high performance on handwriting recognition under sufficient training data sets,and the additional technical optimization greatly enhances the generalization ability and robustness of the model.

关 键 词：手写字符 识别 神经网络 MNIST TensorFlow 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]