DeepLearning-L8-ResNet

2015年，Kaiming He在《Deep Residual Learning for Image Recognition》提出ResNet，将网络深度提升到152层，夺得 ILSVRC 2015的冠军。

1. 深度网络的问题

深层网络能够表示非常复杂的函数，在反向传播过程中，梯度会逐渐消失（假如采用Sigmoid函数，对于幅度为1的信号，每向后传递一层，梯度就衰减为原来的0.25，层数越多，衰减越厉害），导致无法对前面网络层的权重进行有效的调整。

随着网络层级的不断增加，模型精度不断得到提升，而当网络层级增加到一定的数目以后，训练精度和测试精度迅速下降，这说明当网络变得很深以后，深度网络就变得更加难以训练了，因此并不是网络越深越好。

2. ResNet基本模块

ResNets使用"shortcut"或者叫"skip connection" ，使得梯度可以直接传播到前几层。

（1）The identity block

输入输出维度一致

Identity block. Skip connection "skips over" 2 layers.

Identity block. Skip connection "skips over" 3 layers.

First componentof main path:

• The first CONV2D has $F_1$ filters of shape (1,1) and a stride of (1,1). Its padding is “valid” and its name should be conv_name_base + '2a'. Use 0 as the seed for the random initialization.
• The first BatchNorm is normalizing the channels axis. Its name should be bn_name_base + '2a'.
• Then apply the ReLU activation function. This has no name and no hyperparameters.

Second component of main path:

• The second CONV2D has $F_2$ filters of shape $(f,f)$ and a stride of (1,1). Its padding is “same” and its name should be conv_name_base + '2b'. Use 0 as the seed for the random initialization.
• The second BatchNorm is normalizing the channels axis. Its name should be bn_name_base + '2b'.
• Then apply the ReLU activation function. This has no name and no hyperparameters.

Third component of main path:

• The third CONV2D has $F_3$ filters of shape (1,1) and a stride of (1,1). Its padding is “valid” and its name should be conv_name_base + '2c'. Use 0 as the seed for the random initialization.
• The third BatchNorm is normalizing the channels axis. Its name should be bn_name_base + '2c'. Note that there is no ReLU activation function in this component.

Final step:

• The shortcut and the input are added together.
• Then apply the ReLU activation function. This has no name and no hyperparameters.

（2）The convolutional block

输入输出维度不一致

First component of main path:

• The first CONV2D has $F_1$ filters of shape (1,1) and a stride of (s,s). Its padding is “valid” and its name should be conv_name_base + '2a'.
• The first BatchNorm is normalizing the channels axis. Its name should be bn_name_base + '2a'.
• Then apply the ReLU activation function. This has no name and no hyperparameters.

Second component of main path:

• The second CONV2D has $F_2$ filters of (f,f) and a stride of (1,1). Its padding is “same” and it’s name should be conv_name_base + '2b'.
• The second BatchNorm is normalizing the channels axis. Its name should be bn_name_base + '2b'.
• Then apply the ReLU activation function. This has no name and no hyperparameters.

Third component of main path:

• The third CONV2D has $F_3$ filters of (1,1) and a stride of (1,1). Its padding is “valid” and it’s name should be conv_name_base + '2c'.
• The third BatchNorm is normalizing the channels axis. Its name should be bn_name_base + '2c'. Note that there is no ReLU activation function in this component.

Shortcut path:

• The CONV2D has $F_3$ filters of shape (1,1) and a stride of (s,s). Its padding is “valid” and its name should be conv_name_base + '1'.
• The BatchNorm is normalizing the channels axis. Its name should be bn_name_base + '1'.

Final step:

• The shortcut and the main path values are added together.
• Then apply the ReLU activation function. This has no name and no hyperparameters.

3. ResNet网络结构

ResNet-50 model

• Zero-padding pads the input with a pad of (3,3)
• Stage 1:
  • The 2D Convolution has 64 filters of shape (7,7) and uses a stride of (2,2). Its name is “conv1”.
  • BatchNorm is applied to the channels axis of the input.
  • MaxPooling uses a (3,3) window and a (2,2) stride.
    -** Stage 2**:
  • The convolutional block uses three set of filters of size [64,64,256], “f” is 3, “s” is 1 and the block is “a”.
  • The 2 identity blocks use three set of filters of size [64,64,256], “f” is 3 and the blocks are “b” and “c”.
• Stage 3:
  • The convolutional block uses three set of filters of size [128,128,512], “f” is 3, “s” is 2 and the block is “a”.
  • The 3 identity blocks use three set of filters of size [128,128,512], “f” is 3 and the blocks are “b”, “c” and “d”.
• Stage 4:
  • The convolutional block uses three set of filters of size [256, 256, 1024], “f” is 3, “s” is 2 and the block is “a”.
  • The 5 identity blocks use three set of filters of size [256, 256, 1024], “f” is 3 and the blocks are “b”, “c”, “d”, “e” and “f”.
• Stage 5:
  • The convolutional block uses three set of filters of size [512, 512, 2048], “f” is 3, “s” is 2 and the block is “a”.
  • The 2 identity blocks use three set of filters of size [512, 512, 2048], “f” is 3 and the blocks are “b” and “c”.
• The 2D Average Pooling uses a window of shape (2,2) and its name is “avg_pool”.
• The flatten doesn’t have any hyperparameters or name.
• The Fully Connected (Dense) layer reduces its input to the number of classes using a softmax activation. Its name should be 'fc' + str(classes).

《Identity Mappings in Deep Residual Networks》提出了ResNet V2。通过研究 ResNet 残差学习单元的传播公式，发现前馈和反馈信号可以直接传输，因此“shortcut connection”（捷径连接）的非线性激活函数（如ReLU）替换为 Identity Mappings。同时，ResNet V2 在每一层中都使用了 Batch Normalization。这样处理后，新的残差学习单元比以前更容易训练且泛化性更强。