Day51<\/p>\n
import torch
\nimport torch.nn as nn
\nimport torch.optim as optim
\nimport torchvision
\nimport torchvision.transforms as transforms
\nimport matplotlib.pyplot as plt
\nimport numpy as np<\/p>\n
# \u8bbe\u7f6e\u968f\u673a\u79cd\u5b50\u4ee5\u786e\u4fdd\u7ed3\u679c\u53ef\u590d\u73b0
\ntorch.manual_seed(42)<\/p>\n
# \u5b9a\u4e49\u6570\u636e\u9884\u5904\u7406
\ntransform = transforms.Compose([
\n transforms.ToTensor(),
\n transforms.Normalize((0.1307,), (0.3081,))
\n])<\/p>\n
# \u52a0\u8f7dMNIST\u6570\u636e\u96c6
\ntrain_dataset = torchvision.datasets.MNIST(
\n root='.\/data', train=True, download=True, transform=transform)
\ntest_dataset = torchvision.datasets.MNIST(
\n root='.\/data', train=False, download=True, transform=transform)<\/p>\n
# \u521b\u5efa\u6570\u636e\u52a0\u8f7d\u5668
\ntrain_loader = torch.utils.data.DataLoader(
\n train_dataset, batch_size=64, shuffle=True)
\ntest_loader = torch.utils.data.DataLoader(
\n test_dataset, batch_size=1000, shuffle=False)<\/p>\n
# \u5b9a\u4e49CNN\u6a21\u578b
\nclass SimpleCNN(nn.Module):
\n def __init__(self):
\n super(SimpleCNN, self).__init__()
\n self.conv1 = nn.Conv2d(1, 32, kernel_size=3, stride=1, padding=1)
\n self.relu1 = nn.ReLU()
\n self.pool1 = nn.MaxPool2d(kernel_size=2)
\n self.conv2 = nn.Conv2d(32, 64, kernel_size=3, stride=1, padding=1)
\n self.relu2 = nn.ReLU()
\n self.pool2 = nn.MaxPool2d(kernel_size=2)
\n self.fc1 = nn.Linear(64 * 7 * 7, 128)
\n self.relu3 = nn.ReLU()
\n self.fc2 = nn.Linear(128, 10)<\/p>\n
def forward(self, x):
\n x = self.pool1(self.relu1(self.conv1(x)))
\n x = self.pool2(self.relu2(self.conv2(x)))
\n x = x.view(-1, 64 * 7 * 7)
\n x = self.relu3(self.fc1(x))
\n x = self.fc2(x)
\n return x<\/p>\n
# \u521d\u59cb\u5316\u6a21\u578b\u3001\u635f\u5931\u51fd\u6570\u548c\u4f18\u5316\u5668
\ndevice = torch.device("cuda" if torch.cuda.is_available() else "cpu")
\nmodel = SimpleCNN().to(device)
\ncriterion = nn.CrossEntropyLoss()
\noptimizer = optim.Adam(model.parameters(), lr=0.001)<\/p>\n
# \u8bad\u7ec3\u6a21\u578b
\ndef train(epoch):
\n model.train()
\n train_loss = 0
\n correct = 0
\n total = 0<\/p>\n
for batch_idx, (inputs, targets) in enumerate(train_loader):
\n inputs, targets = inputs.to(device), targets.to(device)<\/p>\n
optimizer.zero_grad()
\n outputs = model(inputs)
\n loss = criterion(outputs, targets)
\n loss.backward()
\n optimizer.step()<\/p>\n
train_loss += loss.item()
\n _, predicted = outputs.max(1)
\n total += targets.size(0)
\n correct += predicted.eq(targets).sum().item()<\/p>\n
if batch_idx % 100 == 0:
\n print(f'Epoch: {epoch} | Batch: {batch_idx} | Loss: {train_loss\/(batch_idx+1):.3f} | Acc: {100.*correct\/total:.3f}%')<\/p>\n
return train_loss \/ len(train_loader), 100. * correct \/ total<\/p>\n
# \u6d4b\u8bd5\u6a21\u578b
\ndef test():
\n model.eval()
\n test_loss = 0
\n correct = 0
\n total = 0<\/p>\n
with torch.no_grad():
\n for batch_idx, (inputs, targets) in enumerate(test_loader):
\n inputs, targets = inputs.to(device), targets.to(device)
\n outputs = model(inputs)
\n loss = criterion(outputs, targets)<\/p>\n
test_loss += loss.item()
\n _, predicted = outputs.max(1)
\n total += targets.size(0)
\n correct += predicted.eq(targets).sum().item()<\/p>\n
print(f'Test Loss: {test_loss\/len(test_loader):.3f} | Test Acc: {100.*correct\/total:.3f}%')
\n return test_loss \/ len(test_loader), 100. * correct \/ total<\/p>\n
# \u8bad\u7ec3\u548c\u6d4b\u8bd5\u6a21\u578b
\nepochs = 5
\ntrain_losses, train_accs = [], []
\ntest_losses, test_accs = [], []<\/p>\n
for epoch in range(1, epochs + 1):
\n train_loss, train_acc = train(epoch)
\n test_loss, test_acc = test()<\/p>\n
train_losses.append(train_loss)
\n train_accs.append(train_acc)
\n test_losses.append(test_loss)
\n test_accs.append(test_acc)<\/p>\n
# \u7ed8\u5236\u8bad\u7ec3\u548c\u6d4b\u8bd5\u66f2\u7ebf
\nplt.figure(figsize=(12, 5))
\nplt.subplot(1, 2, 1)
\nplt.plot(train_losses, label='Train Loss')
\nplt.plot(test_losses, label='Test Loss')
\nplt.xlabel('Epoch')
\nplt.ylabel('Loss')
\nplt.legend()
\nplt.title('Loss Curves')<\/p>\n
plt.subplot(1, 2, 2)
\nplt.plot(train_accs, label='Train Accuracy')
\nplt.plot(test_accs, label='Test Accuracy')
\nplt.xlabel('Epoch')
\nplt.ylabel('Accuracy (%)')
\nplt.legend()
\nplt.title('Accuracy Curves')<\/p>\n
plt.tight_layout()
\nplt.savefig('cnn_training_curves.png')
\nplt.show()<\/p>\n
# \u53ef\u89c6\u5316\u4e00\u4e9b\u6d4b\u8bd5\u7ed3\u679c
\ndef visualize_results():
\n model.eval()
\n samples = next(iter(test_loader))
\n images, labels = samples[0][:10].to(device), samples[1][:10]<\/p>\n
with torch.no_grad():
\n outputs = model(images)
\n _, predicted = torch.max(outputs, 1)<\/p>\n
fig = plt.figure(figsize=(12, 5))
\n for i in range(10):
\n ax = fig.add_subplot(2, 5, i+1, xticks=[], yticks=[])
\n ax.imshow(images[i].cpu().squeeze(), cmap='gray')
\n ax.set_title(f'Pred: {predicted[i].item()}, True: {labels[i].item()}',
\n color=("green" if predicted[i] == labels[i] else "red"))<\/p>\n
plt.tight_layout()
\n plt.savefig('cnn_prediction_examples.png')
\n plt.show()<\/p>\n
visualize_results()<\/p>\n
@\u6d59\u5927\u758f\u9526\u884c<\/p>\n","protected":false},"excerpt":{"rendered":"
\u6587\u7ae0\u6d4f\u89c8\u9605\u8bfb269\u6b21\u3002\u3010\u4ee3\u7801\u3011\u3010\u65e0\u6807\u9898\u3011<\/p>\n","protected":false},"author":2,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[4027],"topic":[],"class_list":["post-43450","post","type-post","status-publish","format-standard","hentry","category-server","tag-cocoa"],"yoast_head":"\n