视频链接:动手学习深度学习–网络中的网络(NiN)
课程主页:https://courses.d2l.ai/zh-v2/
教材:https://zh-v2.d2l.ai/
1、NiN网络
NiN架构
总结:
- NiN块使用卷积层加两个1×1卷积层,后者对每个像素增加了非线性性
- NiN使用全局平均池化层来替代VGG和AlexNet中的全连接层
- 不容易过拟合,更少的参数个数
import torch
from torch import nn
from d2l import torch as d2l
import os
os.environ["KMP_DUPLICATE_LIB_OK"]="TRUE"
"====================1、NiN块结构===================="
def nin_block(in_channels, out_channels, kernel_size, strides, padding):
return nn.Sequential(
nn.Conv2d(in_channels, out_channels, kernel_size, strides, padding),
nn.ReLU(),
# 两个1×1卷积层代替全连接层,不会改变通道数
nn.Conv2d(out_channels, out_channels, kernel_size=1), nn.ReLU(),
nn.Conv2d(out_channels, out_channels, kernel_size=1), nn.ReLU())
2、NiN模型
"====================2、NiN模型===================="
net = nn.Sequential(
# 输入通道数96,卷积核11×11,步长为4,边缘填充为0
nin_block(1, 96, kernel_size=11, strides=4, padding=0),
# 最大池化层卷积核3×3,步长为2
nn.MaxPool2d(3, stride=2),
nin_block(96, 256, kernel_size=5, strides=1, padding=2),
nn.MaxPool2d(3, stride=2),
nin_block(256, 384, kernel_size=3, strides=1, padding=1),
nn.MaxPool2d(3, stride=2),
nn.Dropout(0.5),
# 标签类别数是10,所以输出通道数为10
nin_block(384, 10, kernel_size=3, strides=1, padding=1),
# 全局平均池化层
nn.AdaptiveAvgPool2d((1, 1)),
# 将四维的输出转成二维的输出,其形状为(批量大小,10)
nn.Flatten())
# 查看每个块的输出形状
X = torch.rand(size=(1, 1, 224, 224))
for layer in net:
X = layer(X)
print(layer.__class__.__name__,'output shape:\t', X.shape)
'''
输出:
Sequential output shape: torch.Size([1, 96, 54, 54])
MaxPool2d output shape: torch.Size([1, 96, 26, 26])
Sequential output shape: torch.Size([1, 256, 26, 26])
MaxPool2d output shape: torch.Size([1, 256, 12, 12])
Sequential output shape: torch.Size([1, 384, 12, 12])
MaxPool2d output shape: torch.Size([1, 384, 5, 5])
Dropout output shape: torch.Size([1, 384, 5, 5])
Sequential output shape: torch.Size([1, 10, 5, 5])
AdaptiveAvgPool2d output shape: torch.Size([1, 10, 1, 1])
Flatten output shape: torch.Size([1, 10])
'''
3、训练模型
"====================3、训练模型===================="
# 使用Fashion-MNIST来训练模型。训练NiN与训练AlexNet、VGG时相似。
lr, num_epochs, batch_size = 0.1, 10, 128
train_iter, test_iter = d2l.load_data_fashion_mnist(batch_size, resize=224)
d2l.train_ch6(net, train_iter, test_iter, num_epochs, lr, d2l.try_gpu())