基于卷积神经网络的手写数字识别研究与实现

A Research and Application of Handwritten Digit Recognition Based on Convolutional Neural Networks

文章针对计算机视觉领域的手写数字识别问题,介绍了神经网络原理、卷积神经网络(Convolutional Neural Networks,CNN)构成、TensorFlow框架等相关知识,并应用卷积神经网络在MNIST数据集上进行验证。实验结果表明,基于卷积神经网络识别手写数字具有较高的准确性。

This paper targets at the Handwritten Digit Recognition which matters in the computer visual field.This paper introduces the principle and construction of neutral networks as well as the TensorFlow Framework,and it tests the application of Convolutional Neural Networks(CNN)in the MNIST data set.The findings show that Convolutional Neural Networks are conducive to Handwritten Digit Recognition.