12.深度学习练习：Residual Networks（注定成为经典）

本文节选自吴恩达老师《深度学习专项课程》编程作业，在此表示感谢。

课程链接：https://www.deeplearning.ai/deep-learning-specialization/

1 - The problem of very deep neural networks

2 - Building a Residual Network

2.1 - The identity block

2.2 - The convolutional block

3 - Building your first ResNet model (50 layers)

References

Welcome to the second assignment of this week! You will learn how to build very deep convolutional networks, using Residual Networks (ResNets). In theory, very deep networks can represent very complex functions; but in practice, they are hard to train. Residual Networks, introduced by He et al., allow you to train much deeper networks than were previously practically feasible.

In this assignment, you will:

Implement the basic building blocks of ResNets.
Put together these building blocks to implement and train a state-of-the-art neural network for image classification.
This assignment will be done in Keras.

Before jumping into the problem, let's run the cell below to load the required packages.

import numpy as np
import tensorflow as tf
from keras import layers
from keras.layers import Input, Add, Dense, Activation, ZeroPadding2D, BatchNormalization, Flatten, Conv2D, AveragePooling2D, MaxPooling2D, GlobalMaxPooling2D
from keras.models import Model, load_model
from keras.preprocessing import image
from keras.utils import layer_utils
from keras.utils.data_utils import get_file
from keras.applications.imagenet_utils import preprocess_input
import pydot
from IPython.display import SVG
from keras.utils.vis_utils import model_to_dot
from keras.utils import plot_model
from resnets_utils import *
from keras.initializers import glorot_uniform
import scipy.misc
from matplotlib.pyplot import imshow
%matplotlib inlineimport keras.backend as K
K.set_image_data_format('channels_last')
K.set_learning_phase(1)

1 - The problem of very deep neural networks

Last week, you built your first convolutional neural network. In recent years, neural networks have become deeper, with state-of-the-art networks going from just a few layers (e.g., AlexNet) to over a hundred layers.

The main benefit of a very deep network is that it can represent very complex functions. It can also learn features at many different levels of abstraction, from edges (at the lower layers) to very complex features (at the deeper layers). However, using a deeper network doesn't always help. A huge barrier to training them is vanishing gradients: very deep networks often have a gradient signal that goes to zero quickly, thus making gradient descent unbearably slow. More specifically, during gradient descent, as you backprop from the final layer back to the first layer, you are multiplying by the weight matrix on each step, and thus the gradient can decrease exponentially quickly to zero (or, in rare cases, grow exponentially quickly and "explode" to take very large values).

During training, you might therefore see the magnitude (or norm) of the gradient for the earlier layers descrease to zero very rapidly as training proceeds:

2 - Building a Residual Network

In ResNets, a "shortcut" or a "skip connection" allows the gradient to be directly backpropagated to earlier layers:

The image on the left shows the "main path" through the network. The image on the right adds a shortcut to the main path. By stacking these ResNet blocks on top of each other, you can form a very deep network.

We also saw in lecture that having ResNet blocks with the shortcut also makes it very easy for one of the blocks to learn an identity function. This means that you can stack on additional ResNet blocks with little risk of harming training set performance. (There is also some evidence that the ease of learning an identity function--even more than skip connections helping with vanishing gradients--accounts for ResNets' remarkable performance.)

Two main types of blocks are used in a ResNet, depending mainly on whether the input/output dimensions are same or different. You are going to implement both of them.

2.1 - The identity block

The identity block is the standard block used in ResNets, and corresponds to the case where the input activation (say $a^{[l]}$ ) has the same dimension as the output activation (say $a^{[l+2]}$ ). To flesh out the different steps of what happens in a ResNet's identity block, here is an alternative diagram showing the individual steps:

The upper path is the "shortcut path." The lower path is the "main path." In this diagram, we have also made explicit the CONV2D and ReLU steps in each layer. To speed up training we have also added a BatchNorm step. Don't worry about this being complicated to implement--you'll see that BatchNorm is just one line of code in Keras!

In this exercise, you'll actually implement a slightly more powerful version of this identity block, in which the skip connection "skips over" 3 hidden layers rather than 2 layers. It looks like this:

Here're the individual steps.

First component of 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 (?,?) 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.

Exercise: Implement the ResNet identity block. We have implemented the first component of the main path. Please read over this carefully to make sure you understand what it is doing. You should implement the rest.

To implement the Conv2D step: See reference
To implement BatchNorm: See reference (axis: Integer, the axis that should be normalized (typically the channels axis))
For the activation, use: Activation('relu')(X)
To add the value passed forward by the shortcut: See reference

def identity_block(X, f, filters, stage, block):"""Implementation of the identity block as defined in Figure 4Arguments:X -- input tensor of shape (m, n_H_prev, n_W_prev, n_C_prev)f -- integer, specifying the shape of the middle CONV's window for the main pathfilters -- python list of integers, defining the number of filters in the CONV layers of the main pathstage -- integer, used to name the layers, depending on their position in the networkblock -- string/character, used to name the layers, depending on their position in the networkReturns:X -- output of the identity block, tensor of shape (n_H, n_W, n_C)"""# defining name basisconv_name_base = 'res' + str(stage) + block + '_branch'bn_name_base = 'bn' + str(stage) + block + '_branch'# Retrieve FiltersF1, F2, F3 = filters# Save the input value. You'll need this later to add back to the main path. X_shortcut = X# First component of main pathX = Conv2D(filters = F1, kernel_size = (1, 1), strides = (1,1), padding = 'valid', name = conv_name_base + '2a', kernel_initializer = glorot_uniform(seed=0))(X)X = BatchNormalization(axis = 3, name = bn_name_base + '2a')(X)X = Activation('relu')(X)# Second component of main path (≈3 lines)X = Conv2D(filters=F2, kernel_size = (f, f), strides=(1,1), padding='same', name=conv_name_base+'2b', kernel_initializer=glorot_uniform(seed=0))(X)X = BatchNormalization(axis=3, name=bn_name_base + '2b')(X)X= Activation('relu')(X)# Third component of main path (≈2 lines)X = Conv2D(filters = F3, kernel_size = (1, 1), strides = (1,1), padding = 'valid', name = conv_name_base + '2c', kernel_initializer = glorot_uniform(seed=0))(X)X = BatchNormalization(axis=3, name = bn_name_base + '2c')(X)# Final step: Add shortcut value to main path, and pass it through a RELU activation (≈2 lines)X = layers.add([X, X_shortcut])X = Activation('relu')(X)return X

2.2 - The convolutional block

You've implemented the ResNet identity block. Next, the ResNet "convolutional block" is the other type of block. You can use this type of block when the input and output dimensions don't match up. The difference with the identity block is that there is a CONV2D layer in the shortcut path:

The CONV2D layer in the shortcut path is used to resize the input ? to a different dimension, so that the dimensions match up in the final addition needed to add the shortcut value back to the main path. (This plays a similar role as the matrix ??

discussed in lecture.) For example, to reduce the activation dimensions's height and width by a factor of 2, you can use a 1x1 convolution with a stride of 2. The CONV2D layer on the shortcut path does not use any non-linear activation function. Its main role is to just apply a (learned) linear function that reduces the dimension of the input, so that the dimensions match up for the later addition step.

The details of the convolutional block are as follows.

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.

Exercise: Implement the convolutional block. We have implemented the first component of the main path; you should implement the rest. As before, always use 0 as the seed for the random initialization, to ensure consistency with our grader.

Conv Hint
BatchNorm Hint (axis: Integer, the axis that should be normalized (typically the features axis))
For the activation, use: Activation('relu')(X)
Addition Hint

def convolutional_block(X, f, filters, stage, block, s = 2):"""Implementation of the convolutional block as defined in Figure 4Arguments:X -- input tensor of shape (m, n_H_prev, n_W_prev, n_C_prev)f -- integer, specifying the shape of the middle CONV's window for the main pathfilters -- python list of integers, defining the number of filters in the CONV layers of the main pathstage -- integer, used to name the layers, depending on their position in the networkblock -- string/character, used to name the layers, depending on their position in the networks -- Integer, specifying the stride to be usedReturns:X -- output of the convolutional block, tensor of shape (n_H, n_W, n_C)"""# defining name basisconv_name_base = 'res' + str(stage) + block + '_branch'bn_name_base = 'bn' + str(stage) + block + '_branch'# Retrieve FiltersF1, F2, F3 = filters# Save the input valueX_shortcut = X##### MAIN PATH ###### First component of main path X = Conv2D(F1, (1, 1), strides = (s,s), name = conv_name_base + '2a', padding='valid', kernel_initializer = glorot_uniform(seed=0))(X)X = BatchNormalization(axis = 3, name = bn_name_base + '2a')(X)X = Activation('relu')(X)# Second component of main path (≈3 lines)X = Conv2D(F2, (f, f), strides = (1, 1), name = conv_name_base + '2b',padding='same', kernel_initializer = glorot_uniform(seed=0))(X)X = BatchNormalization(axis = 3, name = bn_name_base + '2b')(X)X = Activation('relu')(X)# Third component of main path (≈2 lines)X = Conv2D(F3, (1, 1), strides = (1, 1), name = conv_name_base + '2c',padding='valid', kernel_initializer = glorot_uniform(seed=0))(X)X = BatchNormalization(axis = 3, name = bn_name_base + '2c')(X)##### SHORTCUT PATH #### (≈2 lines)X_shortcut = Conv2D(F3, (1,1), strides=(s,s), name=conv_name_base+'1', padding='valid', kernel_initializer=glorot_uniform(seed=0))(X_shortcut)X_shortcut = BatchNormalization(axis=3, name = bn_name_base + '1')(X_shortcut)# Final step: Add shortcut value to main path, and pass it through a RELU activation (≈2 lines)X = layers.add([X, X_shortcut])X = Activation('relu')(X)return X

3 - Building your first ResNet model (50 layers)

You now have the necessary blocks to build a very deep ResNet. The following figure describes in detail the architecture of this neural network. "ID BLOCK" in the diagram stands for "Identity block," and "ID BLOCK x3" means you should stack 3 identity blocks together.

The details of this ResNet-50 model are:

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 [256, 256, 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).

Exercise: Implement the ResNet with 50 layers described in the figure above. We have implemented Stages 1 and 2. Please implement the rest. (The syntax for implementing Stages 3-5 should be quite similar to that of Stage 2.) Make sure you follow the naming convention in the text above.

You'll need to use this function:

Average pooling see reference

Here're some other functions we used in the code below:

Conv2D: See reference
BatchNorm: See reference (axis: Integer, the axis that should be normalized (typically the features axis))
Zero padding: See reference
Max pooling: See reference
Fully conected layer: See reference
Addition: See reference

def ResNet50(input_shape = (64, 64, 3), classes = 6):"""Implementation of the popular ResNet50 the following architecture:CONV2D -> BATCHNORM -> RELU -> MAXPOOL -> CONVBLOCK -> IDBLOCK*2 -> CONVBLOCK -> IDBLOCK*3-> CONVBLOCK -> IDBLOCK*5 -> CONVBLOCK -> IDBLOCK*2 -> AVGPOOL -> TOPLAYERArguments:input_shape -- shape of the images of the datasetclasses -- integer, number of classesReturns:model -- a Model() instance in Keras"""# Define the input as a tensor with shape input_shapeX_input = Input(input_shape)# Zero-PaddingX = ZeroPadding2D((3, 3))(X_input)# Stage 1X = Conv2D(64, (7, 7), strides = (2, 2), name = 'conv1', kernel_initializer = glorot_uniform(seed=0))(X)X = BatchNormalization(axis = 3, name = 'bn_conv1')(X)X = Activation('relu')(X)X = MaxPooling2D((3, 3), strides=(2, 2))(X)# Stage 2X = convolutional_block(X, f = 3, filters = [64, 64, 256], stage = 2, block='a', s = 1)X = identity_block(X, 3, [64, 64, 256], stage=2, block='b')X = identity_block(X, 3, [64, 64, 256], stage=2, block='c')### START CODE HERE #### Stage 3 (≈4 lines)# 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".X = convolutional_block(X, f=3, filters=[128,128,512], stage=3, block='a', s=2)X = identity_block(X, 3, [128, 128, 512], stage=3, block='b')X = identity_block(X, 3, [128, 128, 512], stage=3, block='c')X = identity_block(X, 3, [128, 128, 512], stage=3, block='d')# Stage 4 (≈6 lines)# 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".X = convolutional_block(X, f = 3, filters=[256, 256, 1024], block='a', stage=4, s = 2)X = identity_block(X, f = 3, filters=[256, 256, 1024], stage=4, block='b')X = identity_block(X, f = 3, filters=[256, 256, 1024], stage=4, block='c')X = identity_block(X, f = 3, filters=[256, 256, 1024], stage=4, block='d')X = identity_block(X, f = 3, filters=[256, 256, 1024], stage=4, block='e')X = identity_block(X, f = 3, filters=[256, 256, 1024], stage=4, block='f')# Stage 5 (≈3 lines)# 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 [256, 256, 2048], "f" is 3 and the blocks are "b" and "c".X = convolutional_block(X, f = 3, filters=[512, 512, 2048], stage=5, block='a', s = 2)# filters should be [256, 256, 2048], but it fail to be graded. Use [512, 512, 2048] to pass the gradingX = identity_block(X, f = 3, filters=[256, 256, 2048], stage=5, block='b')X = identity_block(X, f = 3, filters=[256, 256, 2048], stage=5, block='c')# AVGPOOL (≈1 line). Use "X = AveragePooling2D(...)(X)"# The 2D Average Pooling uses a window of shape (2,2) and its name is "avg_pool".X = AveragePooling2D(pool_size=(2,2))(X)### END CODE HERE #### output layerX = Flatten()(X)X = Dense(classes, activation='softmax', name='fc' + str(classes), kernel_initializer = glorot_uniform(seed=0))(X)# Create modelmodel = Model(inputs = X_input, outputs = X, name='ResNet50')return model

model = ResNet50(input_shape = (64, 64, 3), classes = 6)model.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy'])

Run the following cell to train your model on 2 epochs with a batch size of 32. On a CPU it should take you around 5min per epoch.（注意耗时）

model.fit(X_train, Y_train, epochs = 2, batch_size = 32)

**What you should remember:** - Very deep "plain" networks don't work in practice because they are hard to train due to vanishing gradients. - The skip-connections help to address the Vanishing Gradient problem. They also make it easy for a ResNet block to learn an identity function. - There are two main type of blocks: The identity block and the convolutional block. - Very deep Residual Networks are built by stacking these blocks together.

References

This notebook presents the ResNet algorithm due to He et al. (2015). The implementation here also took significant inspiration and follows the structure given in the github repository of Francois Chollet:

Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun - Deep Residual Learning for Image Recognition (2015)
Francois Chollet's github repository: deep-learning-models/resnet50.py at master · fchollet/deep-learning-models · GitHub