1、有时候在卷积神经网络中进行数据训练，并不要神经网络层越多越复杂性能就越好，可能会由于过拟合。梯度消失也是一种重要的原因，在链式法则将一连串的梯度（如果 梯度<1）乘起来，梯度就会趋于0，权重得不到有效的更新（如下图的 Plain net）。而为了解决这个问题，残差网络出现了，如下简单的残差网络（Residual net）结构：
上图中残差网络 （Residual net）比纯连接网络（Plain net）多了一个跳连接。求梯度的时候，Plain net的：d[H(x)] / dx；Residual net的：d[H(x)] / dx = d[F(x)] / dx + 1。因此，如果遇到了 d[F(x)] / x 非常小于1时，但是总梯度多了1，若干个相乘就很好解决梯度消失的问题。
构建一个简单的卷积残差神经网络：
一个简单残差神经网络的代码：
构造残差网络：
构造残差网络的完整代码：
import torch
from torchvision import transforms
from torchvision import datasets
from torch.utils.data import DataLoader
import torch.nn.functional as F
import torch.optim as optim
batch_size = 64
transform = transforms.Compose([
transforms.ToTensor(), # 将PIL格式图像转换成Tensor矩阵向量（维度28x28转换成1x28x28，1：为RGB通道）【 [0...255]--->[0,1] 】
transforms.Normalize((0.1307, ), (0.3081, )) # 均一化处理（均值、标准差）
])
# 训练集数据
train_dataset = datasets.MNIST(root='../dataset/mnist/',
train=True,
download=True,
transform=transform)
# 加载训练集数据
train_loader = DataLoader(train_dataset,
shuffle=True,
batch_size=batch_size)
# 测试集数据集
test_dataset = datasets.MNIST(root='../dataset/mnist/',
train=False,
download=True,
transform=transform)
# 加载测试集数据集
test_loader = DataLoader(test_dataset,
shuffle=False,
batch_size=batch_size)
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)
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)
class Net(torch.nn.Module):
def __init__(self):
super(Net, self).__init__()
self.conv1 = torch.nn.Conv2d(1, 16, kernel_size=5)
self.conv2 = torch.nn.Conv2d(16, 32, kernel_size=5)
self.mp = torch.nn.MaxPool2d(2)
self.rblock1 = ResidualBlock(16)
self.rblock2 = ResidualBlock(32)
self.fc = torch.nn.Linear(512, 10)
def forward(self, x):
in_size = x.size(0)
x = self.mp(F.relu(self.conv1(x)))
x = self.rblock1(x)
x = self.mp(F.relu(self.conv2(x)))
x = self.rblock2(x)
x = x.view(in_size, -1)
x = self.fc(x)
return x
model = Net()
# 将模型放到GPU上运行，需要加如下两行代码（训练集、测试集中的输入值、实际值也需要加载到GPU上）
# device = torch.device("cude:0" if torch.cuda.is_available() else "cpu")
# model.to(device)
###################3 构建损失函数、优化器###############################
criterion = torch.nn.CrossEntropyLoss() # 交叉熵损失
optimizer = optim.SGD(model.parameters(), lr=0.01, momentum=0.5) # 参数优化
#####################4 循环训练 #########################
def train(epoch):
running_loss = 0.0
for batch_idx, data in enumerate(train_loader, 0):
# 准备数据（input：输入，target：实际值）
inputs, target = data
# 将输入、实际值加载到GPU
# inputs, target = inputs.to(device), target.to(device)
# 梯度清0
optimizer.zero_grad()
# 前向传播
outputs = model(inputs)
# 交叉熵损失函数计算
loss = criterion(outputs, target)
# 反向传播
loss.backward()
# 参数优化
optimizer.step()
# 累计loss
running_loss += loss.item()
# 数据集一共有batch_idx个数据，每隔300个打印一次平均损失函数值
if batch_idx % 300 ==299:
print('[%d, %5d] loss: %.3f' % (epoch + 1, batch_idx + 1, running_loss / 300))
running_loss = 0.0
def test():
correct = 0
total = 0
with torch.no_grad():
for data in test_loader:
images, labels = data
# 将输入、实际值加载到GPU中
# inputs, target = inputs.to(device), target.to(device)
outputs = model(images)
_, predicted = torch.max(outputs.data, dim=1)
total += labels.size(0)
correct += (predicted == labels).sum().item()
print('Accuracy on test set: %d %%' % (100 * correct / total))
if __name__ == '__main__':
for epoch in range(10):
train(epoch)
test()
运行结果：
