pytorch卷積神經網絡如何構建

PyTorch是一個基于Python的科學計算庫，它主要用于深度學習研究，特別是神經網絡。在PyTorch中，你可以使用torch.nn模塊來構建卷積神經網絡（CNN）。以下是一個簡單的CNN示例：

import torch
import torch.nn as nn
import torch.optim as optim

class SimpleCNN(nn.Module):
    def __init__(self, num_classes=10):
        super(SimpleCNN, self).__init__()
        
        # 卷積層1
        self.conv1 = nn.Conv2d(in_channels=3, out_channels=16, kernel_size=3, stride=1, padding=1)
        # 激活函數
        self.relu1 = nn.ReLU()
        # 池化層
        self.pool1 = nn.MaxPool2d(kernel_size=2, stride=2)
        
        # 卷積層2
        self.conv2 = nn.Conv2d(in_channels=16, out_channels=32, kernel_size=3, stride=1, padding=1)
        # 激活函數
        self.relu2 = nn.ReLU()
        # 池化層
        self.pool2 = nn.MaxPool2d(kernel_size=2, stride=2)
        
        # 全連接層
        self.fc1 = nn.Linear(in_features=32 * 25 * 25, out_features=1024)
        self.relu3 = nn.ReLU()
        self.dropout = nn.Dropout(0.5)
        # 輸出層
        self.fc2 = nn.Linear(in_features=1024, out_features=num_classes)
        
    def forward(self, x):
        # 通過卷積層和激活函數
        x = self.conv1(x)
        x = self.relu1(x)
        x = self.pool1(x)
        
        # 通過卷積層和激活函數
        x = self.conv2(x)
        x = self.relu2(x)
        x = self.pool2(x)
        
        # 展平特征圖
        x = x.view(x.size(0), -1)
        
        # 通過全連接層和激活函數
        x = self.fc1(x)
        x = self.relu3(x)
        x = self.dropout(x)
        
        # 輸出結果
        x = self.fc2(x)
        
        return x

# 實例化網絡
num_classes = 10
model = SimpleCNN(num_classes)

# 定義損失函數和優化器
criterion = nn.CrossEntropyLoss()
optimizer = optim.Adam(model.parameters(), lr=0.001)

# 訓練網絡
for epoch in range(num_epochs):
    for images, labels in train_loader:
        optimizer.zero_grad()
        
        outputs = model(images)
        loss = criterion(outputs, labels)
        loss.backward()
        optimizer.step()

這個示例中，我們定義了一個簡單的CNN網絡，包含兩個卷積層、兩個池化層和兩個全連接層。你可以根據你的任務和數據集來調整網絡結構。