这个资源使用实现lenet-5的网络结构来MNIST数据集，代码参考了UFLDL上的相关的代码，以及R. B. Palm实现的CNN中的相关代码，为了适应数据集我把lenet-5输入的大小改为了28*28，c3的每一张特征图都与s4的每一张特征图相关，训练的结果可以达到99.1 -The resources for network structure lenet-5 to MNIST data sets, code reference to the relevant code UFLDL on CNN and in the relevant code RB Palm achieve, in order to adapt to the size of the data set I lenet-5 input to the 28* 28, each have a characteristic c3 FIG FIG characteristics associated with each one s4, the results of training can reach 99.1