定义一个卷积网络并用其训练数据:
定义卷积网络:
import torch as t
import torch.nn as nn
import torch.nn.functional as F
#一般把网络中具有可学习参数的层(例如全连接层、卷积层等)放在构造函数__init__()中
#一般把不具有可学习参数的层(Relu、dropout)放在构造函数中,也可不放在构造函数中
class Net(nn.Module):
def __init__(self):
super(Net,self).__init__()
# 定义卷积层以及全连接层等
self.conv1 = nn.Conv2d(1,6,5) #输入通道数为1,输出为6,filter大小为5、输入通道数与输入数据input要匹配
self.conv2 = nn.Conv2d(6,16,5)
self.fc1 = nn.Linear(16*5*5,120)
self.fc2 = nn.Linear(120,84)
self.fc3 = nn.Linear(84,10)
def forward(self,x):
x = F.max_pool2d(F.relu(self.conv1(x)),(2,2))
# conv1 是Conv2d的一个对象,对象(param)这样的使用是因为在Conv2d父类nn.Module实现了__call__方法,使得其可以作为函数使用 eg:self.conv1(x)
x = F.max_pool2d(F.relu(self.conv2(x)),2)
x = x.view(x.size(0),-1)
x = F.relu(self.fc1(x))
x = F.relu(self.fc2(x))
x = self.fc3(x)
return x
net = Net()
print(net)
结果:
Net(
(conv1): Conv2d(1, 6, kernel_size=(5, 5), stride=(1, 1))
(conv2): Conv2d(6, 16, kernel_size=(5, 5), stride=(1, 1))
(fc1): Linear(in_features=400, out_features=120, bias=True)
(fc2): Linear(in_features=120, out_features=84, bias=True)
(fc3): Linear(in_features=84, out_features=10, bias=True)
)
输入数据的处理:
#forward函数的输入和输出都是Variable,只有Variable才具有自动求导功能,而Tensor是没有的,所以在输入时,需把Tensor封装成Variable。
input = Variable(t.randn(1, 1, 32, 32))
定义损失函数:
nn实现了神经网络中大多数的损失函数,例如nn.MSELoss用来计算均方误差,nn.CrossEntropyLoss用来计算交叉熵损失。
output = net(input)
target = Variable(t.arange(0,10))
criterion = nn.MSELoss()
loss = criterion(output, target)
定义优化器:
import torch.optim as optim
#新建一个优化器,指定要调整的参数和学习率
optimizer = optim.SGD(net.parameters(), lr = 0.01)
# 在训练过程中
# 先梯度清零(与net.zero_grad()效果一样)
optimizer.zero_grad()
# 计算损失
output = net(input)
loss = criterion(output, target)
#反向传播
loss.backward()
#更新参数
optimizer.step()
lenet网络
1、首先加载数据,使用dataset、dataloader
import torchvision as tv
import torchvision.transforms as transforms
from torchvision.transforms import ToPILImage
show = ToPILImage() # 可以把Tensor转成Image,方便可视化
# 第一次运行程序torchvision会自动下载CIFAR-10数据集,
# 大约100M,需花费一定的时间,
# 如果已经下载有CIFAR-10,可通过root参数指定
# 定义对数据的预处理
transform = transforms.Compose([
transforms.ToTensor(), # 转为Tensor
transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5)), # 归一化
])
# 训练集
trainset = tv.datasets.CIFAR10(
root='/home/cy/data/',
train=True,
download=True,
transform=transform)
trainloader = t.utils.data.DataLoader(
trainset,
batch_size=4,
shuffle=True,
num_workers=2)
# 测试集
testset = tv.datasets.CIFAR10(
'/home/cy/data/',
train=False,
download=True,
transform=transform)
testloader = t.utils.data.DataLoader(
testset,
batch_size=4,
shuffle=False,
num_workers=2)
classes = ('plane', 'car', 'bird', 'cat',
'deer', 'dog', 'frog', 'horse', 'ship', 'truck')
2、将dataloader转变成可迭代对象,
dataiter = iter(trainloader)
images, labels = dataiter.next() # 返回4张图片及标签
print(' '.join('%11s'%classes[labels[j]] for j in range(4)))
show(tv.utils.make_grid((images+1)/2)).resize((400,100))
3、定义leNet网络
import torch.nn as nn
import torch.nn.functional as F
class Net(nn.Module):
def __init__(self):
super(Net, self).__init__()
self.conv1 = nn.Conv2d(3, 6, 5)
self.conv2 = nn.Conv2d(6, 16, 5)
self.fc1 = nn.Linear(16*5*5, 120)
self.fc2 = nn.Linear(120, 84)
self.fc3 = nn.Linear(84, 10)
def forward(self, x):
x = F.max_pool2d(F.relu(self.conv1(x)), (2, 2))
x = F.max_pool2d(F.relu(self.conv2(x)), 2)
x = x.view(x.size()[0], -1)
x = F.relu(self.fc1(x))
x = F.relu(self.fc2(x))
x = self.fc3(x)
return x
net = Net()
print(net)
4、定义损失函数和优化器(loss和optimizer)
from torch import optim
criterion = nn.CrossEntropyLoss() # 交叉熵损失函数
optimizer = optim.SGD(net.parameters(), lr=0.001, momentum=0.9)
5、训练网络
所有网络的训练流程都是类似的,不断地执行如下流程:
*输入数据
*前向传播+反向传播
*更新参数
t.set_num_threads(8)
for epoch in range(2):
running_loss = 0.0
for i, data in enumerate(trainloader, 0):
# 输入数据
inputs, labels = data
inputs, labels = Variable(inputs), Variable(labels)
# 梯度清零
optimizer.zero_grad()
# forward + backward
outputs = net(inputs)
loss = criterion(outputs, labels)
loss.backward()
# 更新参数
optimizer.step()
# 打印log信息
running_loss += loss.data[0]
if i % 2000 == 1999: # 每2000个batch打印一下训练状态
print('[%d, %5d] loss: %.3f' \
% (epoch+1, i+1, running_loss / 2000))
running_loss = 0.0
print('Finished Training')
6、取一部分测试网络
dataiter = iter(testloader)
images, labels = dataiter.next() # 一个batch返回4张图片
print('实际的label: ', ' '.join(\
'%08s'%classes[labels[j]] for j in range(4)))
show(tv.utils.make_grid(images / 2 - 0.5)).resize((400,100))
# 计算图片在每个类别上的分数
outputs = net(Variable(images))
# 得分最高的那个类
_, predicted = t.max(outputs.data, 1) # 数据、1表示取一行的最大值
print('预测结果: ', ' '.join('%5s'\
% classes[predicted[j]] for j in range(4)))
7、整个测试集测试网络
correct = 0 # 预测正确的图片数
total = 0 # 总共的图片数
for data in testloader:
images, labels = data
outputs = net(Variable(images))
_, predicted = t.max(outputs.data, 1)
total += labels.size(0)
correct += (predicted == labels).sum()
print('10000张测试集中的准确率为: %d %%' % (100 * correct / total))
8、如果用GPU训练,可修改以下几部分:
if t.cuda.is_available():
net.cuda()
images = images.cuda()
labels = labels.cuda()
output = net(Variable(images))
loss= criterion(output,Variable(labels))