多特征融合代码的简单实现_特征层融合代码

import torch
import torch.nn as nn

class MultimodalNet(nn.Module):
    def __init__(self):
        super(MultimodalNet, self).__init__()
        
        # 定义第一个模态的神经网络层
        self.fc1 = nn.Linear(100, 50)
        self.relu1 = nn.ReLU()
        
        # 定义第二个模态的神经网络层
        self.conv1 = nn.Conv2d(1, 32, kernel_size=5)
        self.pool1 = nn.MaxPool2d(kernel_size=2)
        self.conv2 = nn.Conv2d(32, 64, kernel_size=5)
        self.pool2 = nn.MaxPool2d(kernel_size=2)
        self.relu2 = nn.ReLU()
        
        # 定义第三个模态的神经网络层
        self.lstm = nn.LSTM(10, 20, batch_first=True)
        
        # 定义最终的全连接层
        self.fc2 = nn.Linear(1344, 10)
        
    def forward(self, x1, x2, x3):
        # 第一个模态的前向传播
        x1 = self.fc1(x1)
        x1 = self.relu1(x1)
        
        # 第二个模态的前向传播
        x2 = self.conv1(x2)
        x2 = self.pool1(x2)
        x2 = self.conv2(x2)
        x2 = self.pool2(x2)
        x2 = self.relu2(x2)
        
        # 第三个模态的前向传播
        x3, _ = self.lstm(x3)
        
        # 将三个模态的输出拼接在一起
        x = torch.cat([x1, x2.view(x2.size(0), -1), x3.view(x3.size(0), -1)], dim=1)
        
        # 最终的前向传播
        x = self.fc2(x)
        return x

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