[人工智能]动手学深度学习v2-池化层

import torch
from torch import nn
from d2l import torch as d2l

# 实现池化层的正向传播，无步幅，无填充
def pool2d(X, pool_size, mode='max'): # mode:max,avgs,X是输入
  p_h, p_w = pool_size
  Y = torch.zeros((X.shape[0] - p_h + 1, X.shape[1] - p_w + 1))
  for i in range(Y.shape[0]):
    for j in range(Y.shape[1]):
      if mode == 'max':
        Y[i, j] = X[i:i + p_h, j:j + p_w].max()
      elif mode == 'avg':
        Y[i, j] = X[i:i + p_h, j:j + p_w].mean()
return Y

# 验证二维最大池化层的输出
X = torch.tensor([[0.0, 1.0, 2.0],[3.0, 4.0, 5.0],[6.0, 7.0, 8.0]])
pool2d(X,(2, 2))
# 验证平均池化层
pool22(X, (2,2), 'avg')

# 填充和步幅
X = torch.arange(16, dtype=torch.float32).reshape((1, 1, 4, 4))

# 深度学习框架中的步幅与池化窗口的大小相同,意味着两个不同窗口之间没有重叠
pool2d = nn.MaxPool2d(3)
pool2d(X)

# 填充和步幅可以手动设定
pool2d = nn.MaxPool2d(3, padding=1, stride=2)
pool2d(X)

# 设定一个任意大小的矩阵池化窗口，并分别设定填充和步幅的高度和宽度
pool2d = nn.MaxPool2d((2,3), padding=(1,1), stride=(2,3))
pool2d(X)

# 池化层在每个输入通道上单独运算
X = torch.cat((X, X+1), 1)
X

pool2d = nn.MaxPool2d(3, padding=1, stride=2)
pool2d(X)