简介
卷积神经网络(Convolutional Neural Network, CNN)是深度学习技术中极具代表的网络结构之一，在图像处理领域取得了很大的成功，在国际标准的ImageNet数据集上，许多成功的模型都是基于CNN的。
卷积神经网络CNN的结构一般包含这几个层：
输入层：用于数据的输入
卷积层：使用卷积核进行特征提取和特征映射
激励层：由于卷积也是一种线性运算，因此需要增加非线性映射
池化层：进行下采样，对特征图稀疏处理，减少数据运算量。
全连接层：通常在CNN的尾部进行重新拟合，减少特征信息的损失
输出层：用于输出结果
PyTorch实战
本文选用上篇的数据集MNIST手写数字识别实践CNN。
import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim
from torchvision import datasets, transforms
from torch.autograd import Variable
# Training settings
batch_size = 64
# MNIST Dataset
train_dataset = datasets.MNIST(root='./data/',
train=True,
transform=transforms.ToTensor(),
download=True)
test_dataset = datasets.MNIST(root='./data/',
train=False,
transform=transforms.ToTensor())
# Data Loader (Input Pipeline)
train_loader = torch.utils.data.DataLoader(dataset=train_dataset, batch_size=batch_size, shuffle=True)
test_loader = torch.utils.data.DataLoader(dataset=test_dataset, batch_size=batch_size, shuffle=False)
class Net(nn.Module): def __init__(self): super(Net, self).__init__() # 输入1通道，输出10通道，kernel 5*5 self.conv1 = nn.Conv2d(1, 10, kernel_size=5) self.conv2 = nn.Conv2d(10, 20, kernel_size=5) self.mp = nn.MaxPool2d(2) # fully connect self.fc = nn.Linear(320, 10) def forward(self, x): # in_size = 64 in_size = x.size(0) # one batch # x: 64*10*12*12 x = F.relu(self.mp(self.conv1(x))) # x: 64*20*4*4 x = F.relu(self.mp(self.conv2(x))) # x: 64*320 x = x.view(in_size, -1) # flatten the tensor # x: 64*10 x = self.fc(x) return F.log_softmax(x) model = Net() optimizer = optim.SGD(model.parameters(), lr=0.01, momentum=0.5) def train(epoch): for batch_idx, (data, target) in enumerate(train_loader): data, target = Variable(data), Variable(target) optimizer.zero_grad() output = model(data) loss = F.nll_loss(output, target) loss.backward() optimizer.step() if batch_idx % 200 == 0: print('Train Epoch: {} [{}/{} ({:.0f}%)]\tLoss: {:.6f}'.format( epoch, batch_idx * len(data), len(train_loader.dataset), 100. * batch_idx / len(train_loader), loss.data[0])) def test(): test_loss = 0 correct = 0 for data, target in test_loader: data, target = Variable(data, volatile=True), Variable(target) output = model(data) # sum up batch loss test_loss += F.nll_loss(output, target, size_average=False).data[0] # get the index of the max log-probability pred = output.data.max(1, keepdim=True)[1] correct += pred.eq(target.data.view_as(pred)).cpu().sum() test_loss /= len(test_loader.dataset) print('\nTest set: Average loss: {:.4f}, Accuracy: {}/{} ({:.0f}%)\n'.format( test_loss, correct, len(test_loader.dataset), 100. * correct / len(test_loader.dataset))) for epoch in range(1, 10): train(epoch) test() 输出结果： Train Epoch: 1 [0/60000 (0%)] Loss: 2.315724 Train Epoch: 1 [12800/60000 (21%)] Loss: 1.931551 Train Epoch: 1 [25600/60000 (43%)] Loss: 0.733935 Train Epoch: 1 [38400/60000 (64%)] Loss: 0.165043 Train Epoch: 1 [51200/60000 (85%)] Loss: 0.235188 Test set: Average loss: 0.1935, Accuracy: 9421/10000 (94%) Train Epoch: 2 [0/60000 (0%)] Loss: 0.333513 Train Epoch: 2 [12800/60000 (21%)] Loss: 0.163156 Train Epoch: 2 [25600/60000 (43%)] Loss: 0.213840 Train Epoch: 2 [38400/60000 (64%)] Loss: 0.141114 Train Epoch: 2 [51200/60000 (85%)] Loss: 0.128191 Test set: Average loss: 0.1180, Accuracy: 9645/10000 (96%) Train Epoch: 3 [0/60000 (0%)] Loss: 0.206469 Train Epoch: 3 [12800/60000 (21%)] Loss: 0.234443 Train Epoch: 3 [25600/60000 (43%)] Loss: 0.061048 Train Epoch: 3 [38400/60000 (64%)] Loss: 0.192217 Train Epoch: 3 [51200/60000 (85%)] Loss: 0.089190 Test set: Average loss: 0.0938, Accuracy: 9723/10000 (97%) Train Epoch: 4 [0/60000 (0%)] Loss: 0.086325 Train Epoch: 4 [12800/60000 (21%)] Loss: 0.117741 Train Epoch: 4 [25600/60000 (43%)] Loss: 0.188178 Train Epoch: 4 [38400/60000 (64%)] Loss: 0.049807 Train Epoch: 4 [51200/60000 (85%)] Loss: 0.174097 Test set: Average loss: 0.0743, Accuracy: 9767/10000 (98%) Train Epoch: 5 [0/60000 (0%)] Loss: 0.063171 Train Epoch: 5 [12800/60000 (21%)] Loss: 0.061265 Train Epoch: 5 [25600/60000 (43%)] Loss: 0.103549 Train Epoch: 5 [38400/60000 (64%)] Loss: 0.019137 Train Epoch: 5 [51200/60000 (85%)] Loss: 0.067103 Test set: Average loss: 0.0720, Accuracy: 9781/10000 (98%) Train Epoch: 6 [0/60000 (0%)] Loss: 0.069251 Train Epoch: 6 [12800/60000 (21%)] Loss: 0.075502 Train Epoch: 6 [25600/60000 (43%)] Loss: 0.052337 Train Epoch: 6 [38400/60000 (64%)] Loss: 0.015375 Train Epoch: 6 [51200/60000 (85%)] Loss: 0.028996 Test set: Average loss: 0.0694, Accuracy: 9783/10000 (98%) Train Epoch: 7 [0/60000 (0%)] Loss: 0.171613 Train Epoch: 7 [12800/60000 (21%)] Loss: 0.078520 Train Epoch: 7 [25600/60000 (43%)] Loss: 0.149186 Train Epoch: 7 [38400/60000 (64%)] Loss: 0.026692 Train Epoch: 7 [51200/60000 (85%)] Loss: 0.108824 Test set: Average loss: 0.0672, Accuracy: 9793/10000 (98%) Train Epoch: 8 [0/60000 (0%)] Loss: 0.029188 Train Epoch: 8 [12800/60000 (21%)] Loss: 0.031202 Train Epoch: 8 [25600/60000 (43%)] Loss: 0.194858 Train Epoch: 8 [38400/60000 (64%)] Loss: 0.051497 Train Epoch: 8 [51200/60000 (85%)] Loss: 0.024832 Test set: Average loss: 0.0535, Accuracy: 9837/10000 (98%) Train Epoch: 9 [0/60000 (0%)] Loss: 0.026706 Train Epoch: 9 [12800/60000 (21%)] Loss: 0.057807 Train Epoch: 9 [25600/60000 (43%)] Loss: 0.065225 Train Epoch: 9 [38400/60000 (64%)] Loss: 0.037004 Train Epoch: 9 [51200/60000 (85%)] Loss: 0.057822 Test set: Average loss: 0.0538, Accuracy: 9829/10000 (98%) Process finished with exit code 0 
京公网安备 11010502036488号