残差网络

残差网络的核心思想是：每个附加层都应该更容易地包含原始函数作为其元素之一。

残差块

串联一个层改变函数类，我们希望扩大函数类，残差块加入快速通道来得到f(x)=x+g(x)的结果

ResNet块

1.高宽减半的ResNet块（步幅2）

2.后接多个高宽不变的ResNet块

ResNet架构

1.类似VGG和GoogLeNet总体架构

2.但替换成ResNet块

总结

残差块使得很深的网络更加容易训练，甚至可以训练一千层的网络

代码实现

ResNet沿用了VGG完整的3X3卷积层设计。 残差块里首先有2个有相同输出通道数的3X3卷积层。 每个卷积层后接一个批量规范化层和ReLU激活函数。 然后我们通过跨层数据通路，跳过这2个卷积运算，将输入直接加在最后的ReLU激活函数前。 这样的设计要求2个卷积层的输出与输入形状一样，从而使它们可以相加。 如果想改变通道数，就需要引入一个额外的
1X1卷积层来将输入变换成需要的形状后再做相加运算。

import torch
from torch import nn
from torch.nn import functional as F
from d2l import torch as d2l


class Residual(nn.Module):  #save
    def __init__(self, input_channels, num_channels,
                 use_1x1conv=False, strides=1):
        super().__init__()
        self.conv1 = nn.Conv2d(input_channels, num_channels,
                               kernel_size=3, padding=1, stride=strides)
        self.conv2 = nn.Conv2d(num_channels, num_channels,
                               kernel_size=3, padding=1)
        if use_1x1conv:
            self.conv3 = nn.Conv2d(input_channels, num_channels,
                                   kernel_size=1, stride=strides)
        else:
            self.conv3 = None
        self.bn1 = nn.BatchNorm2d(num_channels)
        self.bn2 = nn.BatchNorm2d(num_channels)

    def forward(self, X):
        Y = F.relu(self.bn1(self.conv1(X)))
        Y = self.bn2(self.conv2(Y))
        if self.conv3:
            X = self.conv3(X)
        Y += X
        return F.relu(Y)

一种是当use_1x1conv=False时，应用ReLU非线性函数之前，将输入添加到输出。 另一种是当use_1x1conv=True时，添加通过1X1卷积调整通道和分辨率。

输入与输出形状一致

blk = Residual(3,3)
X = torch.rand(4, 3, 6, 6)
Y = blk(X)
Y.shape

torch.Size([4, 3, 6, 6])

增加输出通道的同时，减半高和宽

blk = Residual(3,6, use_1x1conv=True, strides=2)
blk(X).shape # batch size, channel, h, w

torch.Size([4, 6, 3, 3])

ResNet模型

b1 = nn.Sequential(nn.Conv2d(1, 64, kernel_size=7, stride=2, padding=3),
                   nn.BatchNorm2d(64), nn.ReLU(),
                   nn.MaxPool2d(kernel_size=3, stride=2, padding=1))

def resnet_block(input_channels, num_channels, num_residuals,
                 first_block=False):
    blk = []
    for i in range(num_residuals):
        if i == 0 and not first_block:
            blk.append(Residual(input_channels, num_channels,
                                use_1x1conv=True, strides=2))
        else:
            blk.append(Residual(num_channels, num_channels))
    return blk

b2 = nn.Sequential(*resnet_block(64, 64, 2, first_block=True))
b3 = nn.Sequential(*resnet_block(64, 128, 2))
b4 = nn.Sequential(*resnet_block(128, 256, 2))
b5 = nn.Sequential(*resnet_block(256, 512, 2))

net = nn.Sequential(b1, b2, b3, b4, b5,
                    nn.AdaptiveAvgPool2d((1,1)),
                    nn.Flatten(), nn.Linear(512, 10))

观察ResNet的不同模块的输入形状是如何变化。

X = torch.rand(size=(1, 1, 224, 224))
for layer in net:
    X = layer(X)
    print(layer.__class__.__name__,'output shape:\t', X.shape)

Sequential output shape:	 torch.Size([1, 64, 56, 56])
Sequential output shape:	 torch.Size([1, 64, 56, 56])
Sequential output shape:	 torch.Size([1, 128, 28, 28])
Sequential output shape:	 torch.Size([1, 256, 14, 14])
Sequential output shape:	 torch.Size([1, 512, 7, 7])
AdaptiveAvgPool2d output shape:	 torch.Size([1, 512, 1, 1])
Flatten output shape:	 torch.Size([1, 512])
Linear output shape:	 torch.Size([1, 10])

训练模型

lr, num_epochs, batch_size = 0.05, 10, 256
train_iter, test_iter = d2l.load_data_fashion_mnist(batch_size, resize=96)
d2l.train_ch6(net, train_iter, test_iter, num_epochs, lr, d2l.try_gpu())

loss 0.016, train acc 0.995, test acc 0.915
1553.6 examples/sec on cuda:0