### PyTorch 实现 MNIST 手写数字识别 以下是基于所提供的引用以及专业知识构建的一个完整的 PyTorch 实现 MNIST 数据集的手写数字识别教程。 #### 加载数据集 通过 `torchvision.datasets` 和 `DataLoader` 来加载并处理 MNIST 数据集。这一步骤中，会定义批量大小 (`batch_size`) 并将数据划分为训练集和测试集[^2]。 ```python import torch from torchvision import datasets, transforms from torch.utils.data import DataLoader # 定义数据预处理操作 transform = transforms.Compose([ transforms.ToTensor(), # 将图片转换为张量 transforms.Normalize((0.5,), (0.5,)) # 归一化 ]) # 下载并加载MNIST数据集 train_dataset = datasets.MNIST(root='./data', train=True, download=True, transform=transform) test_dataset = datasets.MNIST(root='./data', train=False, download=True, transform=transform) # 使用Dataloader加载数据 batch_size = 64 train_loader = DataLoader(dataset=train_dataset, batch_size=batch_size, shuffle=True) test_loader = DataLoader(dataset=test_dataset, batch_size=batch_size, shuffle=False) ``` --- #### 构建 CNN 模型 这里采用了一个简单的卷积神经网络 (CNN)，适合初学者理解和实践[^1]。 ```python import torch.nn as nn import torch.nn.functional as F class Net(nn.Module): def __init__(self): super(Net, self).__init__() self.conv1 = nn.Conv2d(in_channels=1, out_channels=32, kernel_size=3) # 卷积层 self.conv2 = nn.Conv2d(in_channels=32, out_channels=64, kernel_size=3) # 卷积层 self.fc1 = nn.Linear(1600, 128) # 全连接层 self.fc2 = nn.Linear(128, 10) # 输出层 def forward(self, x): x = F.relu(self.conv1(x)) x = F.max_pool2d(x, 2) x = F.relu(self.conv2(x)) x = F.max_pool2d(x, 2) x = torch.flatten(x, 1) # 展平成向量 x = F.relu(self.fc1(x)) x = self.fc2(x) return F.log_softmax(x, dim=1) # 应用log softmax激活函数 ``` --- #### 训练过程 在这一部分，我们将模型送入 GPU 或 CPU 进行训练，并记录损失值以便后续可视化分析[^4]。 ```python device = torch.device("cuda" if torch.cuda.is_available() else "cpu") # 自动检测设备 model = Net().to(device) criterion = nn.NLLLoss() # 负对数似然损失 optimizer = torch.optim.Adam(model.parameters(), lr=0.001) # Adam优化器 def train(epoch): model.train() for batch_idx, (data, target) in enumerate(train_loader): data, target = data.to(device), target.to(device) optimizer.zero_grad() output = model(data) loss = criterion(output, target) loss.backward() optimizer.step() if batch_idx % 100 == 0: print(f'Train Epoch: {epoch} [{batch_idx * len(data)}/{len(train_loader.dataset)} ' f'({100. * batch_idx / len(train_loader):.0f}%)]\tLoss: {loss.item():.6f}') ``` --- #### 测试过程 评估模型性能时，计算每轮 epoch 的平均精度。 ```python def test(): model.eval() test_loss = 0 correct = 0 with torch.no_grad(): for data, target in test_loader: data, target = data.to(device), target.to(device) output = model(data) test_loss += criterion(output, target).item() pred = output.argmax(dim=1, keepdim=True) correct += pred.eq(target.view_as(pred)).sum().item() test_loss /= len(test_loader.dataset) accuracy = 100. * correct / len(test_loader.dataset) print(f'\nTest set: Average loss: {test_loss:.4f}, Accuracy: {correct}/{len(test_loader.dataset)} ({accuracy:.0f}%)\n') ``` --- #### 可视化训练进度 绘制训练损失曲线和测试损失曲线以观察模型的学习情况。 ```python import matplotlib.pyplot as plt train_losses = [] train_counter = [] test_losses = [] test_counter = [] def plot_progress(): fig = plt.figure() plt.plot(train_counter, train_losses, color='blue') plt.scatter(test_counter, test_losses, color='red') plt.legend(['Train Loss', 'Test Loss'], loc='upper right') plt.xlabel('Number of Training Examples Seen') plt.ylabel('Negative Log Likelihood Loss') plt.show() ``` 调用上述方法即可完成整个流程的实现。 ---