- 🍨 本文为🔗365天深度学习训练营 中的学习记录博客
- 🍖 原作者:K同学啊 | 接辅导、项目定制
配置环境
import torch
import torch.nn as nn
import matplotlib.pyplot as plt
import torchvision
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
device
device(type='cuda')
下载训练集和测试集
train_ds = torchvision.datasets.CIFAR10('data2',
train=True,
transform=torchvision.transforms.ToTensor(), # 将数据类型转化为Tensor
download=True)
test_ds = torchvision.datasets.CIFAR10('data2',
train=False,
transform=torchvision.transforms.ToTensor(), # 将数据类型转化为Tensor
download=True)
Downloading https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz to data2\cifar-10-python.tar.gz
100%|██████████| 170498071/170498071 [00:19<00:00, 8536534.91it/s]
Extracting data2\cifar-10-python.tar.gz to data2
Files already downloaded and verified
数据可视化
batch_size = 32
train_dl = torch.utils.data.DataLoader(train_ds,
batch_size=batch_size,
shuffle=True)
test_dl = torch.utils.data.DataLoader(test_ds,
batch_size=batch_size)
# 取一个批次查看数据格式
# 数据的shape为:[batch_size, channel, height, weight]
# 其中batch_size为自己设定,channel,height和weight分别是图片的通道数,高度和宽度。
imgs, labels = next(iter(train_dl))
imgs.shape
torch.Size([32, 3, 32, 32])
import numpy as np
# 指定图片大小,图像大小为20宽、5高的绘图(单位为英寸inch)
plt.figure(figsize=(20, 5))
for i, imgs in enumerate(imgs[:20]):
# 维度缩减
npimg = imgs.numpy().transpose((1, 2, 0))
# 将整个figure分成2行10列,绘制第i+1个子图。
plt.subplot(2, 10, i+1)
plt.imshow(npimg, cmap=plt.cm.binary)
plt.axis('off')
创建模型
import torch.nn.functional as F
num_classes = 10 # 图片的类别数
class Model(nn.Module):
def __init__(self):
super().__init__()
# 特征提取网络
self.conv1 = nn.Conv2d(3, 64, kernel_size=3) # 第一层卷积,卷积核大小为3*3
self.pool1 = nn.MaxPool2d(kernel_size=2) # 设置池化层,池化核大小为2*2
self.conv2 = nn.Conv2d(64, 64, kernel_size=3) # 第二层卷积,卷积核大小为3*3
self.pool2 = nn.MaxPool2d(kernel_size=2)
self.conv3 = nn.Conv2d(64, 128, kernel_size=3) # 第二层卷积,卷积核大小为3*3
self.pool3 = nn.MaxPool2d(kernel_size=2)
# 分类网络
self.fc1 = nn.Linear(512, 256)
self.fc2 = nn.Linear(256, num_classes)
# 前向传播
def forward(self, x):
x = self.pool1(F.relu(self.conv1(x)))
x = self.pool2(F.relu(self.conv2(x)))
x = self.pool3(F.relu(self.conv3(x)))
x = torch.flatten(x, start_dim=1)
x = F.relu(self.fc1(x))
x = self.fc2(x)
return x
torch.nn.Conv2d() 实现了二维卷积操作。在卷积神经网络中,卷积层通过对输入数据应用多个不同的卷积核,并加上偏置项,来提取输入数据的特征。每个卷积核学习从输入数据中提取不同的特征。
查看模型信息
from torchinfo import summary
# 将模型转移到GPU中(我们模型运行均在GPU中进行)
model = Model().to(device)
summary(model)
=================================================================
Layer (type:depth-idx) Param #
=================================================================
Model --
├─Conv2d: 1-1 1,792
├─MaxPool2d: 1-2 --
├─Conv2d: 1-3 36,928
├─MaxPool2d: 1-4 --
├─Conv2d: 1-5 73,856
├─MaxPool2d: 1-6 --
├─Linear: 1-7 131,328
├─Linear: 1-8 2,570
=================================================================
Total params: 246,474
Trainable params: 246,474
Non-trainable params: 0
=================================================================
loss_fn = nn.CrossEntropyLoss() # 创建损失函数
learn_rate = 1e-2 # 学习率
opt = torch.optim.SGD(model.parameters(),lr=learn_rate)
定义训练函数
# 训练循环
def train(dataloader, model, loss_fn, optimizer):
size = len(dataloader.dataset) # 训练集的大小,一共60000张图片
num_batches = len(dataloader) # 批次数目,1875(60000/32)
train_loss, train_acc = 0, 0 # 初始化训练损失和正确率
for X, y in dataloader: # 获取图片及其标签
X, y = X.to(device), y.to(device)
# 计算预测误差
pred = model(X) # 网络输出
loss = loss_fn(pred, y) # 计算网络输出和真实值之间的差距,targets为真实值,计算二者差值即为损失
# 反向传播
optimizer.zero_grad() # grad属性归零
loss.backward() # 反向传播
optimizer.step() # 每一步自动更新
# 记录acc与loss
train_acc += (pred.argmax(1) == y).type(torch.float).sum().item()
train_loss += loss.item()
train_acc /= size
train_loss /= num_batches
return train_acc, train_loss
定义测试函数
# 测试方法
def test (dataloader, model, loss_fn):
size = len(dataloader.dataset) # 测试集的大小,一共10000张图片
num_batches = len(dataloader) # 批次数目,313(10000/32=312.5,向上取整)
test_loss, test_acc = 0, 0
# 当不进行训练时,停止梯度更新,节省计算内存消耗
with torch.no_grad():
for imgs, target in dataloader:
imgs, target = imgs.to(device), target.to(device)
# 计算loss
target_pred = model(imgs)
loss = loss_fn(target_pred, target)
test_loss += loss.item()
test_acc += (target_pred.argmax(1) == target).type(torch.float).sum().item()
test_acc /= size
test_loss /= num_batches
return test_acc, test_loss
开始训练
epochs = 10
train_loss = []
train_acc = []
test_loss = []
test_acc = []
for epoch in range(epochs):
model.train()
epoch_train_acc, epoch_train_loss = train(train_dl, model, loss_fn, opt)
model.eval()
epoch_test_acc, epoch_test_loss = test(test_dl, model, loss_fn)
train_acc.append(epoch_train_acc)
train_loss.append(epoch_train_loss)
test_acc.append(epoch_test_acc)
test_loss.append(epoch_test_loss)
template = ('Epoch:{:2d}, Train_acc:{:.1f}%, Train_loss:{:.3f}, Test_acc:{:.1f}%,Test_loss:{:.3f}')
print(template.format(epoch+1, epoch_train_acc*100, epoch_train_loss, epoch_test_acc*100, epoch_test_loss))
print('Done')
Epoch: 1, Train_acc:15.2%, Train_loss:2.271, Test_acc:18.5%,Test_loss:2.147
Epoch: 2, Train_acc:26.3%, Train_loss:1.993, Test_acc:30.4%,Test_loss:1.938
Epoch: 3, Train_acc:35.7%, Train_loss:1.773, Test_acc:37.3%,Test_loss:1.771
Epoch: 4, Train_acc:41.4%, Train_loss:1.612, Test_acc:43.0%,Test_loss:1.572
Epoch: 5, Train_acc:44.9%, Train_loss:1.515, Test_acc:45.7%,Test_loss:1.523
Epoch: 6, Train_acc:48.2%, Train_loss:1.432, Test_acc:50.1%,Test_loss:1.384
Epoch: 7, Train_acc:51.1%, Train_loss:1.359, Test_acc:47.0%,Test_loss:1.518
Epoch: 8, Train_acc:53.6%, Train_loss:1.294, Test_acc:54.9%,Test_loss:1.274
Epoch: 9, Train_acc:56.0%, Train_loss:1.232, Test_acc:52.7%,Test_loss:1.346
Epoch:10, Train_acc:58.2%, Train_loss:1.180, Test_acc:57.6%,Test_loss:1.190
Done
查看结果
import matplotlib.pyplot as plt
#隐藏警告
import warnings
warnings.filterwarnings("ignore") #忽略警告信息
plt.rcParams['font.sans-serif'] = ['SimHei'] # 用来正常显示中文标签
plt.rcParams['axes.unicode_minus'] = False # 用来正常显示负号
plt.rcParams['figure.dpi'] = 100 #分辨率
epochs_range = range(epochs)
plt.figure(figsize=(12, 3))
plt.subplot(1, 2, 1)
plt.plot(epochs_range, train_acc, label='Training Accuracy')
plt.plot(epochs_range, test_acc, label='Test Accuracy')
plt.legend(loc='lower right')
plt.title('Training and Validation Accuracy')
plt.subplot(1, 2, 2)
plt.plot(epochs_range, train_loss, label='Training Loss')
plt.plot(epochs_range, test_loss, label='Test Loss')
plt.legend(loc='upper right')
plt.title('Training and Validation Loss')
plt.show()
-
训练精度(Training Accuracy)是指模型在训练数据集上进行训练时所达到的准确率,它衡量了模型在已知标签的训练样本上的预测准确程度。
-
测试精度(Test Accuracy)是指模型在独立的测试数据集上进行验证时所达到的准确率,它衡量了模型在未见过的样本上的预测准确程度。测试精度常用来评估模型的泛化能力,即模型对于未知数据的预测能力。
jupyter文件转md文件命令:jupyter nbconvert --to markdown file_path.ipynb
卷积层与池化层的计算
1. 卷积层的计算过程
一个 4 × 4 4 \times 4 4×4 的输入矩阵 I I I,和一个 3 × 3 3 \times 3 3×3 的卷积核 K K K,卷积操作的步骤如下:
输入矩阵
I
I
I:
[
i
11
i
12
i
13
i
14
i
21
i
22
i
23
i
24
i
31
i
32
i
33
i
34
i
41
i
42
i
43
i
44
]
\begin{bmatrix} i_{11} & i_{12} & i_{13} & i_{14} \\ i_{21} & i_{22} & i_{23} & i_{24} \\ i_{31} & i_{32} & i_{33} & i_{34} \\ i_{41} & i_{42} & i_{43} & i_{44} \end{bmatrix}
i11i21i31i41i12i22i32i42i13i23i33i43i14i24i34i44
卷积核
K
K
K:
[
k
11
k
12
k
13
k
21
k
22
k
23
k
31
k
32
k
33
]
\begin{bmatrix} k_{11} & k_{12} & k_{13} \\ k_{21} & k_{22} & k_{23} \\ k_{31} & k_{32} & k_{33} \end{bmatrix}
k11k21k31k12k22k32k13k23k33
进行卷积操作(这里假设步长为1,无填充):
输出 O O O 的第一个元素 o 11 o_{11} o11 通过取输入矩阵的左上角 3 × 3 3 \times 3 3×3 子矩阵与卷积核 K K K 进行元素对应相乘后求和得到:
o 11 = i 11 k 11 + i 12 k 12 + i 13 k 13 + i 21 k 21 + i 22 k 22 + i 23 k 23 + i 31 k 31 + i 32 k 32 + i 33 k 33 o_{11} = i_{11}k_{11} + i_{12}k_{12} + i_{13}k_{13} + i_{21}k_{21} + i_{22}k_{22} + i_{23}k_{23} + i_{31}k_{31} + i_{32}k_{32} + i_{33}k_{33} o11=i11k11+i12k12+i13k13+i21k21+i22k22+i23k23+i31k31+i32k32+i33k33
依此类推,可以计算输出矩阵 O O O 中的其他元素。
输出矩阵大小的计算公式是:
输出大小
=
输入大小
−
核大小
+
2
×
填充
步长
+
1
\text{输出大小} = \frac{\text{输入大小} - \text{核大小} + 2 \times \text{填充}}{\text{步长}} + 1
输出大小=步长输入大小−核大小+2×填充+1
2. 池化层的计算过程
池化层通常用于降低数据的空间尺寸,同时保持重要信息。我们来看最常用的池化操作:最大池化。
假设我们有以下 4 × 4 4 \times 4 4×4 矩阵作为输入,并执行 2 × 2 2 \times 2 2×2 的最大池化操作:
[ p 11 p 12 p 13 p 14 p 21 p 22 p 23 p 24 p 31 p 32 p 33 p 34 p 41 p 42 p 43 p 44 ] \begin{bmatrix} p_{11} & p_{12} & p_{13} & p_{14} \\ p_{21} & p_{22} & p_{23} & p_{24} \\ p_{31} & p_{32} & p_{33} & p_{34} \\ p_{41} & p_{42} & p_{43} & p_{44} \end{bmatrix} p11p21p31p41p12p22p32p42p13p23p33p43p14p24p34p44
最大池化会在 2 × 2 2 \times 2 2×2 的子矩阵中选择最大的元素,因此,池化层的输出将是:
[ max ( p 11 , p 12 , p 21 , p 22 ) max ( p 13 , p 14 , p 23 , p 24 ) max ( p 31 , p 32 , p 41 , p 42 ) max ( p 33 , p 34 , p 43 , p 44 ) ] \begin{bmatrix} \max(p_{11}, p_{12}, p_{21}, p_{22}) & \max(p_{13}, p_{14}, p_{23}, p_{24}) \\ \max(p_{31}, p_{32}, p_{41}, p_{42}) & \max(p_{33}, p_{34}, p_{43}, p_{44}) \end{bmatrix} [max(p11,p12,p21,p22)max(p31,p32,p41,p42)max(p13,p14,p23,p24)max(p33,p34,p43,p44)]
个人小结
- 定义了一个包含卷积层和全连接层的神经网络模型。模型中使用了ReLU激活函数和最大池化来增强特征提取能力和减少计算复杂度。
- 熟悉了关于卷积层和池化层计算过程,增加了对模型内部运作的理解。