神经网络的基本骨架

1. nn.Module的使用

• 所有的模型都要继承 Module 类
• 需要重写初始化函数和运算步骤函数

eg：

import torch.nn as nn
import torch.nn.functional as Fclass Model(nn.Module):		# 继承父类Module def __init__(self):		# 重写初始化函数super().__init__()		# 调用父类初始化self.conv1 = nn.Conv2d(1, 20, 5)self.conv2 = nn.Conv2d(20, 20, 5)def forward(self, x):		# 神经网络的运算步骤--前向传播x = F.relu(self.conv1(x))	# x->卷积->非线性return F.relu(self.conv2(x))	# x->卷积->非线性

代码示例：

import torch
from torch import nnclass Kun(nn.Module):def __init__(self):super().__init__()def forward(self, input):output = input+1	# 实现输出加1return outputkun = Kun()
x = torch.tensor(1.0)
output = kun(x)
print(output)   # tensor(2.)

2. 卷积

conv2可选参数

卷积计算过程示意：

import torch# 输入图像（5*5）
input = torch.tensor([[1, 2, 0, 3, 1],[0, 1, 2, 3, 1],[1, 2, 1, 0, 0],[5, 2, 3, 1, 1],[2, 1, 0, 1, 1]])  # 输入tensor数据类型的二维矩阵# 卷积核
kernel = torch.tensor([[1, 2, 1],[0, 1, 0],[2, 1, 0]])print(input.shape)
print(kernel.shape)

torch.Size([5, 5])
torch.Size([3, 3])

如果不调整尺寸会报错：Expected 3D(unbatched) or 4D(batched) input to conv2d, but got input of size: [5, 5]

所以需要调整

input = torch.reshape(input, (1, 1, 5, 5))
kernel = torch.reshape(kernel, (1, 1, 3, 3))

output = F.conv2d(input, kernel, stride=1)
print(output)--------------------------------------------------------------------------
tensor([[[[10, 12, 12],[18, 16, 16],[13,  9,  3]]]])

stride可以选择移动的步长

output2 = F.conv2d(input, kernel, stride=2)
print(output2)
----------------------------------------------------------------------------
tensor([[[[10, 12],[13,  3]]]])

padding进行填充(默认填充0)

output3 = F.conv2d(input, kernel, stride=1, padding=1)
print(output3)
-----------------------------------------------------------------------------
tensor([[[[ 1,  3,  4, 10,  8],[ 5, 10, 12, 12,  6],[ 7, 18, 16, 16,  8],[11, 13,  9,  3,  4],[14, 13,  9,  7,  4]]]])

示例代码：

import torch
import torch.nn.functional as F
# 输入图像（5*5）
input = torch.tensor([[1, 2, 0, 3, 1],[0, 1, 2, 3, 1],[1, 2, 1, 0, 0],[5, 2, 3, 1, 1],[2, 1, 0, 1, 1]])  # 输入tensor数据类型的二维矩阵# 卷积核
kernel = torch.tensor([[1, 2, 1],[0, 1, 0],[2, 1, 0]])
# 调整输入的尺寸
# 如果不调整尺寸会报错
# Expected 3D(unbatched) or 4D(batched) input to conv2d, but got input of size: [5, 5]
input = torch.reshape(input, (1, 1, 5, 5))
kernel = torch.reshape(kernel, (1, 1, 3, 3))
# print(input.shape)    # torch.Size([1, 1, 5, 5])
# print(kernel.shape)   # torch.Size([1, 1, 3, 3])output = F.conv2d(input, kernel, stride=1)
print(output)output2 = F.conv2d(input, kernel, stride=2)
print(output2)output3 = F.conv2d(input, kernel, stride=1, padding=1)
print(output3)