构建卷积神经网络

• 卷积网络中的输入和层与传统神经网络有些区别，需重新设计，训练模块基本一致

  import torch
  import torch.nn as nn
  import torch.optim as optim
  import torch.nn.functional as F
  from torchvision import datasets,transforms 
  import matplotlib.pyplot as plt
  import numpy as np
  %matplotlib inline

  首先读取数据

• 分别构建训练集和测试集（验证集）
• DataLoader来迭代取数据

  # 定义超参数 
  input_size = 28  #图像的总尺寸28*28
  num_classes = 10  #标签的种类数
  num_epochs = 3  #训练的总循环周期
  batch_size = 64  #一个撮（批次）的大小，64张图片# 训练集
  train_dataset = datasets.MNIST(root='./data',  train=True,   transform=transforms.ToTensor(),  download=True) # 测试集
  test_dataset = datasets.MNIST(root='./data', train=False, transform=transforms.ToTensor())# 构建batch数据
  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=True)

  卷积网络模块构建

• 一般卷积层，relu层，池化层可以写成一个套餐
• 注意卷积最后结果还是一个特征图，需要把图转换成向量才能做分类或者回归任务

  class CNN(nn.Module):def __init__(self):super(CNN, self).__init__()self.conv1 = nn.Sequential(         # 输入大小 (1, 28, 28)nn.Conv2d(in_channels=1,              # 灰度图out_channels=16,            # 要得到几多少个特征图kernel_size=5,              # 卷积核大小stride=1,                   # 步长padding=2,                  # 如果希望卷积后大小跟原来一样，需要设置padding=(kernel_size-1)/2 if stride=1),                              # 输出的特征图为 (16, 28, 28)nn.ReLU(),                      # relu层nn.MaxPool2d(kernel_size=2),    # 进行池化操作（2x2 区域）, 输出结果为： (16, 14, 14))self.conv2 = nn.Sequential(         # 下一个套餐的输入 (16, 14, 14)nn.Conv2d(16, 32, 5, 1, 2),     # 输出 (32, 14, 14)nn.ReLU(),                      # relu层nn.Conv2d(32, 32, 5, 1, 2),nn.ReLU(),nn.MaxPool2d(2),                # 输出 (32, 7, 7))self.conv3 = nn.Sequential(         # 下一个套餐的输入 (16, 14, 14)nn.Conv2d(32, 64, 5, 1, 2),     # 输出 (32, 14, 14)nn.ReLU(),             # 输出 (32, 7, 7))self.out = nn.Linear(64 * 7 * 7, 10)   # 全连接层得到的结果def forward(self, x):x = self.conv1(x)x = self.conv2(x)x = self.conv3(x)x = x.view(x.size(0), -1)           # flatten操作，结果为：(batch_size, 32 * 7 * 7)output = self.out(x)return output

  准确率作为评估标准

  def accuracy(predictions, labels):pred = torch.max(predictions.data, 1)[1] rights = pred.eq(labels.data.view_as(pred)).sum() return rights, len(labels) 

  训练网络模型

  # 实例化
  net = CNN() 
  #损失函数
  criterion = nn.CrossEntropyLoss() 
  #优化器
  optimizer = optim.Adam(net.parameters(), lr=0.001) #定义优化器，普通的随机梯度下降算法#开始训练循环
  for epoch in range(num_epochs):#当前epoch的结果保存下来train_rights = [] for batch_idx, (data, target) in enumerate(train_loader):  #针对容器中的每一个批进行循环net.train()                             output = net(data) loss = criterion(output, target) optimizer.zero_grad() loss.backward() optimizer.step() right = accuracy(output, target) train_rights.append(right) if batch_idx % 100 == 0: net.eval() val_rights = [] for (data, target) in test_loader:output = net(data) right = accuracy(output, target) val_rights.append(right)#准确率计算train_r = (sum([tup[0] for tup in train_rights]), sum([tup[1] for tup in train_rights]))val_r = (sum([tup[0] for tup in val_rights]), sum([tup[1] for tup in val_rights]))print('当前epoch: {} [{}/{} ({:.0f}%)]\t损失: {:.6f}\t训练集准确率: {:.2f}%\t测试集正确率: {:.2f}%'.format(epoch, batch_idx * batch_size, len(train_loader.dataset),100. * batch_idx / len(train_loader), loss.data, 100. * train_r[0].numpy() / train_r[1], 100. * val_r[0].numpy() / val_r[1]))

  当前epoch: 0 [0/60000 (0%)]	损失: 2.300918	训练集准确率: 10.94%	测试集正确率: 10.10%
  当前epoch: 0 [6400/60000 (11%)]	损失: 0.204191	训练集准确率: 78.06%	测试集正确率: 93.31%
  当前epoch: 0 [12800/60000 (21%)]	损失: 0.039503	训练集准确率: 86.51%	测试集正确率: 96.69%
  当前epoch: 0 [19200/60000 (32%)]	损失: 0.057866	训练集准确率: 89.93%	测试集正确率: 97.54%
  当前epoch: 0 [25600/60000 (43%)]	损失: 0.069566	训练集准确率: 91.68%	测试集正确率: 97.68%
  当前epoch: 0 [32000/60000 (53%)]	损失: 0.228793	训练集准确率: 92.85%	测试集正确率: 98.18%
  当前epoch: 0 [38400/60000 (64%)]	损失: 0.111003	训练集准确率: 93.72%	测试集正确率: 98.16%
  当前epoch: 0 [44800/60000 (75%)]	损失: 0.110226	训练集准确率: 94.28%	测试集正确率: 98.44%
  当前epoch: 0 [51200/60000 (85%)]	损失: 0.014538	训练集准确率: 94.78%	测试集正确率: 98.60%
  当前epoch: 0 [57600/60000 (96%)]	损失: 0.051019	训练集准确率: 95.14%	测试集正确率: 98.45%
  当前epoch: 1 [0/60000 (0%)]	损失: 0.036383	训练集准确率: 98.44%	测试集正确率: 98.68%
  当前epoch: 1 [6400/60000 (11%)]	损失: 0.088116	训练集准确率: 98.50%	测试集正确率: 98.37%
  当前epoch: 1 [12800/60000 (21%)]	损失: 0.120306	训练集准确率: 98.59%	测试集正确率: 98.97%
  当前epoch: 1 [19200/60000 (32%)]	损失: 0.030676	训练集准确率: 98.63%	测试集正确率: 98.83%
  当前epoch: 1 [25600/60000 (43%)]	损失: 0.068475	训练集准确率: 98.59%	测试集正确率: 98.87%
  当前epoch: 1 [32000/60000 (53%)]	损失: 0.033244	训练集准确率: 98.62%	测试集正确率: 99.03%
  当前epoch: 1 [38400/60000 (64%)]	损失: 0.024162	训练集准确率: 98.67%	测试集正确率: 98.81%
  当前epoch: 1 [44800/60000 (75%)]	损失: 0.006713	训练集准确率: 98.69%	测试集正确率: 98.17%
  当前epoch: 1 [51200/60000 (85%)]	损失: 0.009284	训练集准确率: 98.69%	测试集正确率: 98.97%
  当前epoch: 1 [57600/60000 (96%)]	损失: 0.036536	训练集准确率: 98.68%	测试集正确率: 98.97%
  当前epoch: 2 [0/60000 (0%)]	损失: 0.125235	训练集准确率: 98.44%	测试集正确率: 98.73%
  当前epoch: 2 [6400/60000 (11%)]	损失: 0.028075	训练集准确率: 99.13%	测试集正确率: 99.17%
  当前epoch: 2 [12800/60000 (21%)]	损失: 0.029663	训练集准确率: 99.26%	测试集正确率: 98.39%
  当前epoch: 2 [19200/60000 (32%)]	损失: 0.073855	训练集准确率: 99.20%	测试集正确率: 98.81%
  当前epoch: 2 [25600/60000 (43%)]	损失: 0.018130	训练集准确率: 99.16%	测试集正确率: 99.09%
  当前epoch: 2 [32000/60000 (53%)]	损失: 0.006968	训练集准确率: 99.15%	测试集正确率: 99.11%