Pytorch实战3——GoogLeNet对MNIST数据集分类

目录

一、GoogLeNet网络

二、编程实战对MNIST数据集分类

---

 

一、GoogLeNet网络

 输入分为4个通道，每个通道分别做不同的运算，最后拼接起来

 二、编程实战对MNIST数据集分类

代码基于上一篇的卷积神经网络编程，只需要把上一篇的代码中设计模型部分替换成维GoogLeNet网络的即可，其它部分代码不变，替换的模型如下

# GoogLeNet卷积神经网络模型
class InceptionA(nn.Module): #import torch.nn 就可以直接写(nn.Module)
    def __init__(self, in_channels):
        super(InceptionA, self).__init__()
        self.branch1x1 = nn.Conv2d(in_channels, 16, kernel_size=1)
        self.branch5x5_1 = nn.Conv2d(in_channels,16, kernel_size=1)
        self.branch5x5_2 = nn.Conv2d(16, 24, kernel_size=5, padding=2)
        self.branch3x3_1 = nn.Conv2d(in_channels, 16, kernel_size=1)
        self.branch3x3_2 = nn.Conv2d(16, 24, kernel_size=3, padding=1)
        self.branch3x3_3 = nn.Conv2d(24, 24, kernel_size=3, padding=1)
        self.branch_pool = nn.Conv2d(in_channels, 24, kernel_size=1)
    def forward(self, x):
        branch1x1 = self.branch1x1(x)
        branch5x5 = self.branch5x5_1(x)
        branch5x5 = self.branch5x5_2(branch5x5)
        branch3x3 = self.branch3x3_1(x)
        branch3x3 = self.branch3x3_2(branch3x3)
        branch3x3 = self.branch3x3_3(branch3x3)
        branch_pool = F.avg_pool2d(x, kernel_size=3, stride=1, padding=1)
        branch_pool = self.branch_pool(branch_pool) 
        outputs = [branch1x1, branch5x5, branch3x3, branch_pool] #
        return torch.cat(outputs, dim=1) #输出维度 = 24+24+24+16=88，按channel方向拼接起来
    
class Net(nn.Module):
    def __init__(self):
        super(Net, self).__init__()
        self.conv1 = nn.Conv2d(1, 10, kernel_size=5)
        self.conv2 = nn.Conv2d(88, 20, kernel_size=5)
        self.incep1 = InceptionA(in_channels=10)
        self.incep2 = InceptionA(in_channels=20)
        self.mp = nn.MaxPool2d(2)
        self.fc = nn.Linear(1408, 10)
    def forward(self, x):  # (batchsize，channels，weight，high)
        in_size = x.size(0)# x.size(0)指batchsize的值，
        x = F.relu(self.mp(self.conv1(x)))  # 1*28*28 -> 10*24*24 -> 10*12*12
        x = self.incep1(x)                 # 输出88*12*12
        x = F.relu(self.mp(self.conv2(x))) #88*12*12 - 20*8*8 -20*4*4
        x = self.incep2(x)                # 88*4*4 =1408
        x = x.view(in_size, -1) #(x.size(0),-1)将tensor的结构转换为了(batchsize, channels*w*h)，即将（channels，w，h）拉直
        x = self.fc(x)
        return x
    
model = Net()

运行结果：

[1,   300] loss: 0.842
[1,   600] loss: 0.214
[1,   900] loss: 0.157
Accuracy on test set: 96 %
[2,   300] loss: 0.128
[2,   600] loss: 0.106
[2,   900] loss: 0.089
Accuracy on test set: 97 %

 由于我只训练了2轮，准确率未提升，多训练一些，会发现准确率提高到98%