.. _sec_vgg:
使用块的网络（VGG）
===================
虽然AlexNet证明深层神经网络卓有成效，但它没有提供一个通用的模板来指导后续的研究人员设计新的网络。
在下面的几个章节中，我们将介绍一些常用于设计深层神经网络的启发式概念。
与芯片设计中工程师从放置晶体管到逻辑元件再到逻辑块的过程类似，神经网络架构的设计也逐渐变得更加抽象。研究人员开始从单个神经元的角度思考问题，发展到整个层，现在又转向块，重复层的模式。
使用块的想法首先出现在牛津大学的\ `视觉几何组（visual geometry
group） `__\ 的\ *VGG网络*\ 中。通过使用循环和子程序，可以很容易地在任何现代深度学习框架的代码中实现这些重复的架构。
VGG块
-----
经典卷积神经网络的基本组成部分是下面的这个序列：
1. 带填充以保持分辨率的卷积层；
2. 非线性激活函数，如ReLU；
3. 汇聚层，如最大汇聚层。
而一个VGG块与之类似，由一系列卷积层组成，后面再加上用于空间下采样的最大汇聚层。在最初的VGG论文中
:cite:`Simonyan.Zisserman.2014`\ ，作者使用了带有\ :math:`3\times3`\ 卷积核、填充为1（保持高度和宽度）的卷积层，和带有\ :math:`2 \times 2`\ 汇聚窗口、步幅为2（每个块后的分辨率减半）的最大汇聚层。在下面的代码中，我们定义了一个名为\ ``vgg_block``\ 的函数来实现一个VGG块。
该函数有三个参数，分别对应于卷积层的数量\ ``num_convs``\ 、输入通道的数量\ ``in_channels``
和输出通道的数量\ ``out_channels``.
.. raw:: html
.. raw:: html
.. raw:: latex
\diilbookstyleinputcell
.. code:: python
from mindspore import nn, ops
from d2l import mindspore as d2l
def vgg_block(num_convs, in_channels, out_channels):
layers = []
for _ in range(num_convs):
layers.append(nn.Conv2d(in_channels, out_channels,
kernel_size=3, padding=1, pad_mode='pad',
weight_init='xavier_uniform'))
layers.append(nn.ReLU())
in_channels = out_channels
layers.append(nn.MaxPool2d(kernel_size=2,stride=2))
return nn.SequentialCell(layers)
.. raw:: html
.. raw:: html
.. raw:: latex
\diilbookstyleinputcell
.. code:: python
import torch
from torch import nn
from d2l import torch as d2l
def vgg_block(num_convs, in_channels, out_channels):
layers = []
for _ in range(num_convs):
layers.append(nn.Conv2d(in_channels, out_channels,
kernel_size=3, padding=1))
layers.append(nn.ReLU())
in_channels = out_channels
layers.append(nn.MaxPool2d(kernel_size=2,stride=2))
return nn.Sequential(*layers)
.. raw:: html
.. raw:: html
.. raw:: html
.. raw:: latex
\diilbookstyleinputcell
.. code:: python
def vgg(conv_arch):
conv_blks = []
in_channels = 1
for (num_convs, out_channels) in conv_arch:
conv_blks.append(vgg_block(num_convs, in_channels, out_channels))
in_channels = out_channels
return nn.SequentialCell([
*conv_blks, nn.Flatten(),
nn.Dense(out_channels * 7 * 7, 4096, weight_init='xavier_uniform'), nn.ReLU(), nn.Dropout(keep_prob=1-0.5),
nn.Dense(4096, 4096, weight_init='xavier_uniform'), nn.ReLU(), nn.Dropout(keep_prob=1-0.5),
nn.Dense(4096, 10, weight_init='xavier_uniform')])
net = vgg(conv_arch)
.. raw:: latex
\diilbookstyleoutputcell
.. parsed-literal::
:class: output
[WARNING] ME(2918374:137997219145536,MainProcess):2026-05-21-02:25:59.358.000 [mindspore/nn/layer/basic.py:176] For Dropout, this parameter `keep_prob` will be deprecated, please use `p` instead.
[WARNING] ME(2918374:137997219145536,MainProcess):2026-05-21-02:25:59.373.000 [mindspore/nn/layer/basic.py:176] For Dropout, this parameter `keep_prob` will be deprecated, please use `p` instead.
.. raw:: html
.. raw:: html
.. raw:: latex
\diilbookstyleinputcell
.. code:: python
def vgg(conv_arch):
conv_blks = []
in_channels = 1
# 卷积层部分
for (num_convs, out_channels) in conv_arch:
conv_blks.append(vgg_block(num_convs, in_channels, out_channels))
in_channels = out_channels
return nn.Sequential(
*conv_blks, nn.Flatten(),
# 全连接层部分
nn.Linear(out_channels * 7 * 7, 4096), nn.ReLU(), nn.Dropout(0.5),
nn.Linear(4096, 4096), nn.ReLU(), nn.Dropout(0.5),
nn.Linear(4096, 10))
net = vgg(conv_arch)
.. raw:: html
.. raw:: html
.. raw:: html
.. raw:: latex
\diilbookstyleinputcell
.. code:: python
X = ops.randn(1, 1, 224, 224)
for blk in net:
X = blk(X)
print(blk.__class__.__name__,'output shape:\t',X.shape)
.. raw:: latex
\diilbookstyleoutputcell
.. parsed-literal::
:class: output
SequentialCell output shape: (1, 64, 112, 112)
SequentialCell output shape: (1, 128, 56, 56)
SequentialCell output shape: (1, 256, 28, 28)
SequentialCell output shape: (1, 512, 14, 14)
SequentialCell output shape: (1, 512, 7, 7)
Flatten output shape: (1, 25088)
Dense output shape: (1, 4096)
ReLU output shape: (1, 4096)
Dropout output shape: (1, 4096)
Dense output shape: (1, 4096)
ReLU output shape: (1, 4096)
Dropout output shape: (1, 4096)
Dense output shape: (1, 10)
.. raw:: html
.. raw:: html
.. raw:: latex
\diilbookstyleinputcell
.. code:: python
X = torch.randn(size=(1, 1, 224, 224))
for blk in net:
X = blk(X)
print(blk.__class__.__name__,'output shape:\t',X.shape)
.. raw:: latex
\diilbookstyleoutputcell
.. parsed-literal::
:class: output
Sequential output shape: torch.Size([1, 64, 112, 112])
Sequential output shape: torch.Size([1, 128, 56, 56])
Sequential output shape: torch.Size([1, 256, 28, 28])
Sequential output shape: torch.Size([1, 512, 14, 14])
Sequential output shape: torch.Size([1, 512, 7, 7])
Flatten output shape: torch.Size([1, 25088])
Linear output shape: torch.Size([1, 4096])
ReLU output shape: torch.Size([1, 4096])
Dropout output shape: torch.Size([1, 4096])
Linear output shape: torch.Size([1, 4096])
ReLU output shape: torch.Size([1, 4096])
Dropout output shape: torch.Size([1, 4096])
Linear output shape: torch.Size([1, 10])
.. raw:: html
.. raw:: html
.. raw:: html
.. raw:: latex
\diilbookstyleinputcell
.. code:: python
ratio = 4
small_conv_arch = [(pair[0], pair[1] // ratio) for pair in conv_arch]
net = vgg(small_conv_arch)
.. raw:: latex
\diilbookstyleoutputcell
.. parsed-literal::
:class: output
[WARNING] ME(2918374:137997219145536,MainProcess):2026-05-21-02:26:00.448.000 [mindspore/nn/layer/basic.py:176] For Dropout, this parameter `keep_prob` will be deprecated, please use `p` instead.
[WARNING] ME(2918374:137997219145536,MainProcess):2026-05-21-02:26:00.468.000 [mindspore/nn/layer/basic.py:176] For Dropout, this parameter `keep_prob` will be deprecated, please use `p` instead.
.. raw:: html
.. raw:: html
.. raw:: latex
\diilbookstyleinputcell
.. code:: python
ratio = 4
small_conv_arch = [(pair[0], pair[1] // ratio) for pair in conv_arch]
net = vgg(small_conv_arch)
.. raw:: html
.. raw:: html
.. raw:: html
.. raw:: latex
\diilbookstyleinputcell
.. code:: python
lr, num_epochs, batch_size = 0.05, 3, 128
train_iter, test_iter = d2l.load_data_fashion_mnist(batch_size, resize=224)
d2l.train_ch6(net, train_iter, test_iter, num_epochs, lr)
.. raw:: latex
\diilbookstyleoutputcell
.. parsed-literal::
:class: output
loss 0.342, train acc 0.875, test acc 0.884
4339.2 examples/sec
.. figure:: output_vgg_4a7574_41_1.png
.. raw:: html
.. raw:: html
.. raw:: latex
\diilbookstyleinputcell
.. code:: python
lr, num_epochs, batch_size = 0.05, 10, 128
train_iter, test_iter = d2l.load_data_fashion_mnist(batch_size, resize=224)
d2l.train_ch6(net, train_iter, test_iter, num_epochs, lr, d2l.try_gpu())
.. raw:: latex
\diilbookstyleoutputcell
.. parsed-literal::
:class: output
loss 0.171, train acc 0.937, test acc 0.926
4357.4 examples/sec on cuda:0
.. figure:: output_vgg_4a7574_44_1.svg
.. raw:: html
.. raw:: html
.. raw:: html
`讨论 `__
.. raw:: html
.. raw:: html
`讨论 `__
.. raw:: html
.. raw:: html