PyTorch——NiN网络的实现

参考链接

1. https://tangshusen.me/Dive-into-DL-PyTorch/#/chapter05_CNN/5.8_nin

NiN块

NiN使用1×1卷积层来替代全连接层，从而使空间信息能够自然传递到后面的层中去。NiN块是NiN中的基础块。它由一个卷积层加两个充当全连接层的1×1卷积层串联而成。其中第一个卷积层的超参数可以自行设置，而第二和第三个卷积层的超参数一般是固定的。

def nin_block(in_channels, out_channels, kernel_size, stride, padding):
    blk = nn.Sequential(nn.Conv2d(in_channels, out_channels, kernel_size, stride, padding),
                        nn.ReLU(),
                        nn.Conv2d(out_channels, out_channels, kernel_size=1),
                        nn.ReLU(),
                        nn.Conv2d(out_channels, out_channels, kernel_size=1),
                        nn.ReLU())
    return blk

NiN vs AlexNet

1. NiN使用卷积窗口形状分别为11×11、5×5和3×3的卷积层，相应的输出通道数与AlexNet中的一致。每个NiN块后接一个步幅为2、窗口形状为3×3的最大池化层。

2. NiN去掉了AlexNet最后的3个全连接层，使用了输出通道数等于标签类别数的NiN块，然后使用全局平均池化层对每个通道中所有元素求平均并直接用于分类。

3. NiN的这个设计的好处是可以显著减小模型参数尺寸，从而缓解过拟合。然而，该设计有时会造成获得有效模型的训练时间的增加。

加载数据

使用“Fashion-MNIST”数据集。读取数据的时候我们额外做了一步将图像高和宽扩大到AlexNet使用的图像高和宽224。这个可以通过torchvision.transforms.Resize实例来实现。也就是说，我们在ToTensor实例前使用Resize实例，然后使用Compose实例来将这两个变换串联。

def load_data_fashion_mnist(batch_size, resize=None, root='~/Datasets/FashionMNIST'):
    if sys.platform.startswith('win'):
        num_workers = 0
    else:
        num_workers = 4
    trans = []
    if resize:
        trans.append(torchvision.transforms.Resize(size=resize))
    trans.append(torchvision.transforms.ToTensor())

    transform = torchvision.transforms.Compose(trans)
    mnist_train = torchvision.datasets.FashionMNIST(root=root, train=True, download=True, transform=transform)
    mnist_test = torchvision.datasets.FashionMNIST(root=root, train=False, download=True, transform=transform)

    train_iter = torch.utils.data.DataLoader(mnist_train, batch_size=batch_size, shuffle=True, num_workers=num_workers)
    test_iter = torch.utils.data.DataLoader(mnist_test, batch_size=batch_size, shuffle=False, num_workers=num_workers)

    return train_iter, test_iter

batch_size = 128
train_iter, test_iter = load_data_fashion_mnist(batch_size, resize=224)

构建模型

net = nn.Sequential(
    nin_block(1, 96, kernel_size=11, stride=4, padding=0),
    nn.MaxPool2d(kernel_size=3, stride=2),
    nin_block(96, 256, kernel_size=5, stride=1, padding=2),
    nn.MaxPool2d(kernel_size=3, stride=2),
    nin_block(256, 384, kernel_size=3, stride=1, padding=1),
    nn.MaxPool2d(kernel_size=3, stride=2), 
    nn.Dropout(0.5),
    # 标签类别数是10
    nin_block(384, 10, kernel_size=3, stride=1, padding=1),
    GlobalAvgPool2d(), 
    # 将四维的输出转成二维的输出，其形状为(批量大小, 10)
    FlattenLayer())

损失函数

损失函数使用交叉熵损失。

loss = torch.nn.CrossEntropyLoss()

优化方法

优化方法使用Adam算法。

optimizer = torch.optim.Adam(net.parameters(), lr=lr)

完整代码

import time
import torch
from torch import nn, optim
import torchvision
import sys

device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')

def load_data_fashion_mnist(batch_size, resize=None, root='~/Datasets/FashionMNIST'):
    if sys.platform.startswith('win'):
        num_workers = 0
    else:
        num_workers = 4
    trans = []
    if resize:
        trans.append(torchvision.transforms.Resize(size=resize))
    trans.append(torchvision.transforms.ToTensor())

    transform = torchvision.transforms.Compose(trans)
    mnist_train = torchvision.datasets.FashionMNIST(root=root, train=True, download=True, transform=transform)
    mnist_test = torchvision.datasets.FashionMNIST(root=root, train=False, download=True, transform=transform)

    train_iter = torch.utils.data.DataLoader(mnist_train, batch_size=batch_size, shuffle=True, num_workers=num_workers)
    test_iter = torch.utils.data.DataLoader(mnist_test, batch_size=batch_size, shuffle=False, num_workers=num_workers)

    return train_iter, test_iter

batch_size = 128
train_iter, test_iter = load_data_fashion_mnist(batch_size, resize=224)

def nin_block(in_channels, out_channels, kernel_size, stride, padding):
    blk = nn.Sequential(nn.Conv2d(in_channels, out_channels, kernel_size, stride, padding),
                        nn.ReLU(),
                        nn.Conv2d(out_channels, out_channels, kernel_size=1),
                        nn.ReLU(),
                        nn.Conv2d(out_channels, out_channels, kernel_size=1),
                        nn.ReLU())
    return blk

import torch.nn.functional as F
class GlobalAvgPool2d(nn.Module):
    def __init__(self):
        super(GlobalAvgPool2d, self).__init__()
    def forward(self, x):
        return F.avg_pool2d(x, kernel_size=x.size()[2:])

class FlattenLayer(nn.Module):
    def __init__(self):
        super(FlattenLayer, self).__init__()
    def forward(self, x): # x shape: (batch, *, *, ...)
        return x.view(x.shape[0], -1)
        
net = nn.Sequential(
    nin_block(1, 96, kernel_size=11, stride=4, padding=0),
    nn.MaxPool2d(kernel_size=3, stride=2),
    nin_block(96, 256, kernel_size=5, stride=1, padding=2),
    nn.MaxPool2d(kernel_size=3, stride=2),
    nin_block(256, 384, kernel_size=3, stride=1, padding=1),
    nn.MaxPool2d(kernel_size=3, stride=2), 
    nn.Dropout(0.5),
    # 标签类别数是10
    nin_block(384, 10, kernel_size=3, stride=1, padding=1),
    GlobalAvgPool2d(), 
    # 将四维的输出转成二维的输出，其形状为(批量大小, 10)
    FlattenLayer())

def evaluate_accuracy(data_iter, net, device=None):
    if device is None and isinstance(net, torch.nn.Module):
        # 如果没指定device就使用net的device
        device = list(net.parameters())[0].device
    acc_sum, n = 0.0, 0
    with torch.no_grad():
        for X, y in data_iter:
            net.eval() # 评估模式, 这会关闭dropout
            acc_sum += (net(X.to(device)).argmax(dim=1) == y.to(device)).float().sum().cpu().item()
            net.train() # 改回训练模式
            n += y.shape[0]
    return acc_sum / n


def train(net, train_iter, test_iter, batch_size, optimizer, device, num_epochs):
    net = net.to(device)
    print("training on ", device)
    loss = torch.nn.CrossEntropyLoss()
    for epoch in range(num_epochs):
        train_l_sum, train_acc_sum, n, batch_count, start = 0.0, 0.0, 0, 0, time.time()
        for X, y in train_iter:
            X = X.to(device)
            y = y.to(device)
            y_hat = net(X)
            l = loss(y_hat, y)
            optimizer.zero_grad()
            l.backward()
            optimizer.step()
            train_l_sum += l.cpu().item()
            train_acc_sum += (y_hat.argmax(dim=1) == y).sum().cpu().item()
            n += y.shape[0]
            batch_count += 1
        test_acc = evaluate_accuracy(test_iter, net)
        print('epoch %d, loss %.4f, train acc %.3f, test acc %.3f, time %.1f sec'
              % (epoch + 1, train_l_sum / batch_count, train_acc_sum / n, test_acc, time.time() - start))

lr, num_epochs = 0.001, 5
optimizer = torch.optim.Adam(net.parameters(), lr=lr)
train(net, train_iter, test_iter, batch_size, optimizer, device, num_epochs)