Home 机器学习 PyTorch

思路

导入库

1
2
3
4
5
6
import torch
from torchvision import transforms
from torchvision import datasets
from torch.utils.data import DataLoader
import torch.nn.functional as F
import torch.optim as optim
  • import torch：PyTorch的核心库，提供张量(Tensor)操作、自动微分、GPU加速等功能
  • from torchvision import transforms：提供图像预处理工具，如尺寸调整、归一化、数据增强、
  • from torchvision import datasets：提供预置数据集（如MNIST、CIFAR）的快速下载和加载接口
  • from torch.utils.data import DataLoader：数据加载器，将数据集分批次(batch)加载，支持多线程加速和数据打乱
  • import torch.nn.functional as F：包含无需学习参数的神经网络函数（如激活函数、损失函数）
  • import torch.optim as optim：提供优化算法，如SGD等

准备数据集

1
2
3
4
5
6
7
batch_size = 64
transform = transforms.Compose([transforms.ToTensor(), transforms.Normalize((0,1307,), (0.3801,))])

train_dataset = datasets.MINST(root='../dataset/mnist', train=True, download=True, transform=transform)
train_loader = DataLoader(train_dataset, shuffle=True, batch_size=batch_size)
test_dataset = datasets.MNIST(root='../dataset/mnist', train=False, download=True, transform=transform)
test_loader = DataLoader(test_dataset, shuffle=False, batch_size=batch_size)
  • 0.1307是总像素的均值
  • 0.3081是总像素的标准差

模型定义

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
class Net(torch.nn.Module):
    def __init__(self):
        super(Net, self).__init__()
        self.conv1 = torch.nn.Conv2d(1, 10, kernel_size=5)
        self.conv2 = torch.nn.Conv2d(10, 20, kernel_size=5)
        self.pooling = torch.nn.MaxPool2d(2)
        self.fc = torch.nn.Linear(320, 10)

    def forward(self, x):
        batch_size = x.size(0)
        x = F.relu(self.pooling(self.conv1(x)))
        x = F.relu(self.pooling(self.conv2(x)))
        x = x.view(batch_size, -1)
        x = self.fc(x)

        return x
  • __init__(self)
    • 浅层(conv1)：提取基础特征（如边缘）
    • 深层(conv2)：组合基础特征为高级特征
    • 池化：逐步压缩空间信息，保留重要特征
    • 全连接：将空间特征转换为分类结果
  • forward(self, x)
    • 特征提取：通过两次$卷积\Rightarrow池化\Rightarrow ReLU$组合，逐步提取从简单到复杂的特征
    • 非线性激活：每次卷积后使用ReLU激活函数，增强模型非线性表达能力
    • 维度压缩：池化层减少计算量并增强平移不变性^1
    • 分类输出：全连接层直接输出分类结果

损失函数和优化器

1
2
criterion = torch.nn.CrossEntropyLoss()
optimizer = optim.SGD(model.parameters(), lr=0.01, momentum=0.5)
  • model.parameters()返回模型中所有需要被优化（训练）的参数
    • conv层的权重weight和偏置bias
    • fc层的权重和偏置
  • momentum=0.5：SGD优化器的动量参数

$$
\begin{aligned}
v_{W_i} &:= \gamma v_{W_i} + \eta \frac{\partial L}{\partial W_i} \
W_i &:= W_i - v_{W_i}
\end{aligned}
$$

训练函数

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
def train(epoch):
    running_loss = 0.0
    for batch_idx, data in enumerate(train_loader, 0):
        inputs, target = data
        optimizer.zero_grad()
        
        outputs = model(inputs)
        loss = criterion(outputs, target)
        loss.backward()
        optimizer.step()
        
        running_loss += loss.item()
        if batch_idx % 300 == 299:
            print('[%d, %5d] loss: %.3f' % (epoch + 1, batch_idx + 1, running_loss / 300))
        running_loss = 0.0
  • for batch_idx, data in enumerate(train_loader, 0)
    • train_loader：PyTorch的DataLoader，按批次加载训练数据
    • enumerate(train_loader, 0)
      • 返回批次索引batch_idx（从0开始）和对应批次数据data
    • inputs, target = data
      • inputs是模型输入
      • target是真实标签

代码

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
import torch
from torchvision import transforms
from torchvision import datasets
from torch.utils.data import DataLoader
import torch.nn.functional as F
import torch.optim as optim

batch_size = 64
transform = transforms.Compose([transforms.ToTensor(), transforms.Normalize((0.1307,), (0.3081,))])

train_dataset = datasets.MNIST(root='../dataset/mnist/', train=True, download=True, transform=transform)
train_loader = DataLoader(train_dataset, shuffle=True, batch_size=batch_size)
test_dataset = datasets.MNIST(root='../dataset/mnist/', train=False, download=True, transform=transform)
test_loader = DataLoader(test_dataset, shuffle=True, batch_size=batch_size)

class Net(torch.nn.Module):
    def __init__(self):
        super(Net, self).__init__()
        self.conv1 = torch.nn.Conv2d(1, 10, kernel_size=5)
        self.conv2 = torch.nn.Conv2d(10, 20, kernel_size=5)
        self.pooling = torch.nn.MaxPool2d(2)
        self.fc = torch.nn.Linear(320, 10)

    def forward(self, x):
        batch_size = x.size(0)
        x = F.relu(self.pooling(self.conv1(x)))
        x = F.relu(self.pooling(self.conv2(x)))
        x = x.view(batch_size, -1)
        x = self.fc(x)

        return x

model = Net()
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model.to(device)

criterion = torch.nn.CrossEntropyLoss()
optimizer = optim.SGD(model.parameters(), lr=0.01, momentum=0.5)

def train(epoch):
    running_loss = 0.0
    for batch_idx, data in enumerate(train_loader, 0):
        inputs, target = data
        inputs, target = inputs.to(device), target.to(device)
        optimizer.zero_grad()

        outputs = model(inputs)
        loss = criterion(outputs, target)
        loss.backward()
        optimizer.step()

        running_loss += loss.item()
        if batch_idx % 300 == 299:
            print('[%d, %5d] loss: %.3f' % (epoch + 1, batch_idx + 1, running_loss / 300))
        running_loss = 0.0