当前位置：网站首页>《PyTorch深度学习实践》第十一课（卷积神经网络CNN高级版）

《PyTorch深度学习实践》第十一课（卷积神经网络CNN高级版）

2022-08-05 05:40:00 【falldeep】

b站刘二视频，地址：

《PyTorch深度学习实践》完结合集_哔哩哔哩_bilibili

concatenate将张量合并（试一试不同的路线）

1*1卷积核的作用，减少运算次数

代码实现Inception模块

梯度消失

56层3 * 3 的网络效果反而跑不过20层的，因为在反向传播中，是每一层网络的导数值乘起来，因为每个导数的值都小于1，当有值特别小的时候，总结果就会趋近于0，导致最开始那几层网络得不到训练。

解决方法1

逐层训练，然后训练完一层后上锁

解决方法2 Residual net

加了一个x后，导数永远大于1

Residual block的实现

import torch
import torch.nn.functional as F
class ResidualBlock(torch.nn.Module):
    def __init__(self, channels):
        super(ResidualBlock, self).__init__()
        self.channels = channels
        self.conv1 = torch.nn.Conv2d(channels, channels, kernel_size=3, padding=1) #padding是为了使得输出像素等于输入像素
        self.conv2 = torch.nn.Conv2d(channels, channels, kernel_size=3, padding=1)
        
    def forward(self, x):
        y = F.relu(self.conv1(x))
        y = self.conv2(y)
        return F.relu(x + y) #加上x

深度学习学习路线

作业代码

1.使用inception的代码

import torch
from torchvision import transforms
from torchvision import datasets
from torch.utils.data import DataLoader
import torch.nn as nn
import torch.nn.functional as F
import matplotlib.pyplot as plt

#step1 prepare data
BATCH_SIZE = 64
transform = transforms.Compose([transforms.ToTensor(), transforms.Normalize((0.1307,), (0.3081,))])
train_set = datasets.MNIST(root='mnist', download=False, transform=transform, train=True)
train_loader = DataLoader(train_set, batch_size=BATCH_SIZE, shuffle=True)
test_set = datasets.MNIST(root='mnist', download=False, transform=transform, train=False)
test_loader = DataLoader(test_set, batch_size=BATCH_SIZE, shuffle=False)


#step2 construct network
class inception(nn.Module):
    def __init__(self, in_channels):
        super(inception, self).__init__()

        self.branch1x1 = nn.Conv2d(in_channels, 16, kernel_size=1)

        self.branch5x5_1 = nn.Conv2d(in_channels, 16, kernel_size=1)
        self.branch5x5_2 = nn.Conv2d(16, 24, kernel_size=5, padding=2)

        self.branch3x3_1 = nn.Conv2d(in_channels, 16, kernel_size=1)
        self.branch3x3_2 = nn.Conv2d(16, 24, kernel_size=3, padding=1)
        self.branch3x3_3 = nn.Conv2d(24, 24, kernel_size=3, padding=1)

        self.branch_pool = nn.Conv2d(in_channels, 24, kernel_size=1)

    def forward(self, x):
        branch1x1 = self.branch1x1(x)

        branch5x5 = self.branch5x5_1(x)
        branch5x5 = self.branch5x5_2(branch5x5)

        branch3x3 = self.branch3x3_1(x)
        branch3x3 = self.branch3x3_2(branch3x3)
        branch3x3 = self.branch3x3_3(branch3x3)

        branch_pool = F.avg_pool2d(x, kernel_size=3, stride=1, padding=1)
        branch_pool = self.branch_pool(branch_pool)

        outputs = [branch1x1, branch5x5, branch3x3, branch_pool]
        return torch.cat(outputs, dim=1) #在通道维度合并

class Net(nn.Module):
    def __init__(self):
        super(Net, self).__init__()
        self.conv1 = nn.Conv2d(1, 10, kernel_size=5)
        self.conv2 = nn.Conv2d(88, 20, kernel_size=5)

        self.incep1 = inception(10)
        self.incep2 = inception(20)

        self.mp = nn.MaxPool2d(kernel_size=2)
        self.fc = nn.Linear(1408, 10)

    def forward(self, x):
        in_size = x.size(0)
        x = F.relu(self.mp(self.conv1(x)))
        x = self.incep1(x)
        x = F.relu(self.mp(self.conv2(x)))
        x = self.incep2(x)
        x = x.view(in_size, -1)
        x = self.fc(x)

        return x

net = Net()

#step3 损失函数和优化函数
criterion =  nn.CrossEntropyLoss()
optimizor = torch.optim.SGD(net.parameters(), lr=0.01, momentum=0.5)

def train():
    running_loss = 0.0
    for i, data in enumerate(train_loader, 0):
        inputs, lables = data
        y_pred = net(inputs)
        loss = criterion(y_pred, lables)

        running_loss += loss.item()
        optimizor.zero_grad()
        loss.backward()
        optimizor.step()

        if i % 300 == 299:
            print('[%d, %5d] loss: %.3f' % (epoch + 1, i + 1, running_loss / 300))
            loss_lst.append(running_loss / 300)
            running_loss = 0

def test():
    correct = 0
    total = 0
    with torch.no_grad():
        for i, data in enumerate(test_loader, 0):
            inputs, lables = data
            outputs = net(inputs)
            _, predicted = torch.max(outputs.data, dim=1)
            total += lables.size(0)
            correct += (predicted == lables).sum().item()
        print('accuracy on test set: %d %% ' % (100 * correct / total))
        acc_lst.append(100 * correct / total)

#step4 训练过程
if __name__ == '__main__':
    acc_lst = []
    loss_lst = []

    for epoch in range(10):
        train()
        test()

    num_lst = [i for i in range(len(loss_lst))]
    plt.plot(num_lst, loss_lst)
    plt.xlabel("i")
    plt.ylabel("loss")
    plt.show()

    num_lst = [i for i in range(len(acc_lst))]
    plt.plot(num_lst, acc_lst)
    plt.xlabel("epoch")
    plt.ylabel("accurate_rate")
    plt.show()

2.使用ResidualBlock的代码

import torch
from torchvision import transforms
from torchvision import datasets
import torch.nn as nn
import torch.nn.functional as F
import matplotlib.pyplot as plt
from torch.utils.data import DataLoader

#step1
BATCH_SIZE = 64
transform = transforms.Compose([transforms.ToTensor(), transforms.Normalize((0.1307,), (0.3081,))])
train_set = datasets.MNIST(root='mnist', download=False, transform=transform, train=True)
train_loader = DataLoader(train_set, shuffle=True, batch_size=BATCH_SIZE)
test_set = datasets.MNIST(root='mnist', download=False, transform=transform, train=False)
test_loader = DataLoader(test_set, shuffle=False, batch_size=BATCH_SIZE)

#step2
class ResidualBlock(nn.Module):
    def __init__(self, channels):
        super(ResidualBlock, self).__init__()
        self.conv1 = nn.Conv2d(channels, channels, kernel_size=3, padding=1)
        self.conv2 = nn.Conv2d(channels, channels, kernel_size=3, padding=1)

    def forward(self, x):
        y = F.relu(self.conv1(x))
        y = self.conv1(y)

        return F.relu(y + x)

class Net(nn.Module):
    def __init__(self):
        super(Net, self).__init__()
        self.conv1 = nn.Conv2d(1, 16, kernel_size=5)
        self.conv2 = nn.Conv2d(16, 32, kernel_size=5)

        self.rblock1 = ResidualBlock(16)
        self.rblock2 = ResidualBlock(32)

        self.pool = nn.MaxPool2d(2)
        self.fc = nn.Linear(512, 10)

    def forward(self, x):
        in_size = x.size(0)
        x = F.relu(self.pool(self.conv1(x)))
        x = self.rblock1(x)
        x = F.relu(self.pool(self.conv2(x)))
        x = self.rblock2(x)

        x = x.view(in_size, -1)
        x = self.fc(x)
        return x

net = Net()

#step3
criterion = nn.CrossEntropyLoss()
optimizor = torch.optim.SGD(net.parameters(), lr=0.01, momentum=0.5)

#step4

def train():
    running_loss = 0.0
    for i, data in enumerate(train_loader, 0):
        inputs, lables = data
        y_pred = net(inputs)
        loss = criterion(y_pred, lables)

        optimizor.zero_grad()
        loss.backward()
        optimizor.step()
        running_loss += loss.item()

        if i % 300 == 299:
            print("[%d, %5d]loss: %.3f" %(epoch + 1, i + 1, running_loss / 300))
            loss_lst.append(running_loss / 300)
            running_loss = 0.0


def test():
    correct = 0
    total = 0
    with torch.no_grad():
        for i, data in enumerate(test_loader, 0):
            inputs, lables = data
            outputs = net(inputs)
            _, predicted = torch.max(outputs, dim=1)

            correct += (predicted == lables).sum().item()
            total += lables.size(0)

        print('accuracy on test set: %d %%' %(100 * correct / total))
        acc_lst.append(100 * correct / total)


if __name__ == '__main__':
    acc_lst = []
    loss_lst = []

    for epoch in range(10):
        train()
        test()

    num_lst = [i for i in range(len(loss_lst))]
    plt.plot(num_lst, loss_lst)
    plt.xlabel("i")
    plt.ylabel("loss")
    plt.show()

    num_lst = [i for i in range(len(acc_lst))]
    plt.plot(num_lst, acc_lst)
    plt.xlabel("epoch")
    plt.ylabel("accurate_rate")
    plt.show()