卷积神经网络实现手写数字识别

导入相关的库

import torch
import torch.nn as nn
from torch.autograd import Variable
import torch.utils.data as Data
import torchvision
import matplotlib.pyplot as plt

设置相应的超参数

EPOCH=1
BATCH_SIZE=50
LR=0.001
DOWNLOAD_MNIST=False

加载训练数据

train_data=torchvision.datasets.MNIST(
    root='./mnist',
    train=True,
    transform=torchvision.transforms.ToTensor(),    #将下载的文件转换成pytorch认识的tensor类型，且将图片的数值大小从（0-255）归一化到（0-1）
    download=True
)

任意显示一个图片

print(train_data.data.size())
print(train_data.targets.size())
plt.imshow(train_data.data[0].numpy(),cmap='gray')
plt.title('%i'%train_data.targets[0])
plt.show()

torch.Size([60000, 28, 28])
torch.Size([60000])

设置训练集和测试集

train_loader=Data.DataLoader(dataset=train_data, batch_size=BATCH_SIZE, shuffle=True)
test_data=torchvision.datasets.MNIST(
    root='./mnist',
    train=False,
)

取数据并处理

with torch.no_grad():
    test_x=Variable(torch.unsqueeze(test_data.data, dim=1)).type(torch.FloatTensor)[:2000]/255   #只取前两千个数据吧，差不多已经够用了，然后将其归一化。
    test_y=test_data.targets[:2000]  

卷积网络的定义

class CNN(nn.Module):
    def __init__(self):
        super(CNN,self).__init__()
        self.conv1=nn.Sequential(
            nn.Conv2d(
                in_channels=1,
                out_channels=16,
                kernel_size=5,
                stride=1,
                padding=2,
            ),
            nn.ReLU(),
            nn.MaxPool2d(kernel_size=2),
        )
        self.conv2=nn.Sequential(
            nn.Conv2d(
                in_channels=16,
                out_channels=32,
                kernel_size=5,
                stride=1,
                padding=2,
            ),
            nn.ReLU(),
            nn.MaxPool2d(kernel_size=2),
        )
        self.out=nn.Linear(32*7*7,10)
    def forward(self,x):
        x=self.conv1(x)
        x=self.conv2(x)     #（batch,32,7,7）
        x=x.view(x.size(0),-1)    #(batch ,32 * 7 * 7)
        output=self.out(x)
        return output

卷积神经网络的结构

cnn=CNN()
print(cnn)

CNN(
  (conv1): Sequential(
    (0): Conv2d(1, 16, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2))
    (1): ReLU()
    (2): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
  )
  (conv2): Sequential(
    (0): Conv2d(16, 32, kernel_size=(5, 5), stride=(1, 1), padding=(2, 2))
    (1): ReLU()
    (2): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
  )
  (out): Linear(in_features=1568, out_features=10, bias=True)
)

训练以及预测

optimizer=torch.optim.Adam(cnn.parameters(),lr=LR)
# 指定损失函数使用交叉信息熵
loss_fn=nn.CrossEntropyLoss()

'''
开始训练我们的模型哦
'''
step=0
for epoch in range(EPOCH):
    #加载训练数据
    for step,data in enumerate(train_loader):
        x,y=data
        b_x=Variable(x)
        b_y=Variable(y)
        output=cnn(b_x)
        loss=loss_fn(output,b_y)
        optimizer.zero_grad()
        loss.backward()
        optimizer.step()
        if (step%50==0):
            test_output=cnn(test_x)
            y_pred=torch.max(test_output,1)[1].data.squeeze()
            accuracy=sum(y_pred==test_y).item()/test_y.size(0)
            print('now epoch :  ', epoch, '   |  loss : %.4f ' % loss.item(), '     |   accuracy :   ' , accuracy)
test_output=cnn(test_x[:10])
y_pred=torch.max(test_output,1)[1].data.squeeze()
print(y_pred.tolist(),'predecton Result')
print(test_y[:10].tolist(),'Real Result')

now epoch :   0    |  loss : 2.3108       |   accuracy :    0.18
now epoch :   0    |  loss : 0.5315       |   accuracy :    0.838
now epoch :   0    |  loss : 0.2425       |   accuracy :    0.891
now epoch :   0    |  loss : 0.3464       |   accuracy :    0.921
now epoch :   0    |  loss : 0.2871       |   accuracy :    0.9405
now epoch :   0    |  loss : 0.1521       |   accuracy :    0.937
now epoch :   0    |  loss : 0.1424       |   accuracy :    0.952
now epoch :   0    |  loss : 0.0820       |   accuracy :    0.955
now epoch :   0    |  loss : 0.0958       |   accuracy :    0.9585
now epoch :   0    |  loss : 0.0694       |   accuracy :    0.9615
now epoch :   0    |  loss : 0.1700       |   accuracy :    0.9645
now epoch :   0    |  loss : 0.0583       |   accuracy :    0.9595
now epoch :   0    |  loss : 0.1821       |   accuracy :    0.9645
now epoch :   0    |  loss : 0.0640       |   accuracy :    0.966
now epoch :   0    |  loss : 0.1602       |   accuracy :    0.9735
now epoch :   0    |  loss : 0.0350       |   accuracy :    0.972
now epoch :   0    |  loss : 0.0863       |   accuracy :    0.9585
now epoch :   0    |  loss : 0.1025       |   accuracy :    0.9655
now epoch :   0    |  loss : 0.0898       |   accuracy :    0.972
now epoch :   0    |  loss : 0.0819       |   accuracy :    0.976
now epoch :   0    |  loss : 0.0593       |   accuracy :    0.976
now epoch :   0    |  loss : 0.0615       |   accuracy :    0.9745
now epoch :   0    |  loss : 0.0529       |   accuracy :    0.978
now epoch :   0    |  loss : 0.0372       |   accuracy :    0.977
[7, 2, 1, 0, 4, 1, 4, 9, 5, 9] predecton Result
[7, 2, 1, 0, 4, 1, 4, 9, 5, 9] Real Result