20241008深度学习动手篇

1.如何写一个神经网络进行训练?

1.1创建一个子类,搭建你需要的神经网络结构

# @File: 241008LeNet.py
# @Author: chen_song
# @Time: 2024/10/8 上午8:31import torch
from torch import nn
from d2l import torch as d2lnet = nn.Sequential(# 进行卷积操作以后,nn.Conv2d(1,6,kernel_size=5,padding=2),nn.Sigmoid(),nn.AvgPool2d(2,stride=2),nn.Conv2d(6,16,kernel_size=5),nn.Sigmoid(),nn.AvgPool2d(2,stride=2),nn.Flatten(),nn.Linear(16*5*5,120),nn.Sigmoid(),nn.Linear(120,84),nn.Sigmoid(),nn.Linear(84,10)
)
print(net)
print("===============================")
X = torch.rand(size=(1,1,28,28),dtype=torch.float32)
Y  = X.copy_(X)
for layer in net:X = layer(X)print(layer.__class__.__name__,X.shape)print("============================")
# 输入给定以后,会进行一系列张量乘法计算
A = net(Y)
# print the last result
print(A)

result below:

Sequential( (0): Conv2d(1, 6, kernel_size=(5, 5), stride=(1, 1),
padding=(2, 2)) (1): Sigmoid() (2): AvgPool2d(kernel_size=2,
stride=2, padding=0) (3): Conv2d(6, 16, kernel_size=(5, 5),
stride=(1, 1)) (4): Sigmoid() (5): AvgPool2d(kernel_size=2,
stride=2, padding=0) (6): Flatten(start_dim=1, end_dim=-1) (7):
Linear(in_features=400, out_features=120, bias=True) (8): Sigmoid()
(9): Linear(in_features=120, out_features=84, bias=True) (10):
Sigmoid() (11): Linear(in_features=84, out_features=10, bias=True) )
=============================== Conv2d torch.Size([1, 6, 28, 28]) Sigmoid torch.Size([1, 6, 28, 28]) AvgPool2d torch.Size([1, 6, 14,
14]) Conv2d torch.Size([1, 16, 10, 10]) Sigmoid torch.Size([1, 16, 10,
10]) AvgPool2d torch.Size([1, 16, 5, 5]) Flatten torch.Size([1, 400])
Linear torch.Size([1, 120]) Sigmoid torch.Size([1, 120]) Linear
torch.Size([1, 84]) Sigmoid torch.Size([1, 84]) Linear torch.Size([1,
10])
============================ tensor([[-0.2278, -0.5057, -0.6303, 0.1526, -0.1510, -0.1933, -0.3120, -0.7823,
0.4070, -0.0717]], grad_fn=)

1.2 加载数据集

batch_size = 256
train_iter, test_iter = d2l.load_data_fashion_mnist(batch_size)

打断点调试:

你会发现:
train_iter和test_iter都是一个torch.utils.dataLoader对象,里面包含几个成员变量,住关键的是dataset对象以及sample对象,仔细研究你就会发现,为啥需要数据加载器了,因为你用神经网络进行训练,数据格式总得对吧,再就是要给个label吧,也就是目标值target吧,所以有余力朋友可以自己设计一个数据加载器…

1.3 自定义一些指标评估函数

def evaluate_accuracy_gpu(net, data_iter, device=None):  # @save"""使用GPU计算模型在数据集上的精度"""if isinstance(net, nn.Module):net.eval()  # 设置为评估模式if not device:device = next(iter(net.parameters())).device# 正确预测的数量，总预测的数量metric = d2l.Accumulator(2)with torch.no_grad():for X, y in data_iter:if isinstance(X, list):# BERT微调所需的（之后将介绍）=== 自然语言处理X = [x.to(device) for x in X]else:X = X.to(device)y = y.to(device)metric.add(d2l.accuracy(net(X), y), y.numel())return metric[0] / metric[1]

注意一下里面net.eval()和net.train()

1.4训练

def train_ch6(net, train_iter, test_iter, num_epochs, lr, device):"""用GPU训练模型(在第六章定义)"""def init_weights(m):if type(m) == nn.Linear or type(m) == nn.Conv2d:nn.init.xavier_uniform_(m.weight)net.apply(init_weights)print('training on', device)net.to(device)optimizer = torch.optim.SGD(net.parameters(), lr=lr)loss = nn.CrossEntropyLoss()animator = d2l.Animator(xlabel='epoch', xlim=[1, num_epochs],legend=['train loss', 'train acc', 'test acc'])timer, num_batches = d2l.Timer(), len(train_iter)for epoch in range(num_epochs):# 训练损失之和，训练准确率之和，样本数metric = d2l.Accumulator(3)net.train()for i, (X, y) in enumerate(train_iter):timer.start()optimizer.zero_grad()X, y = X.to(device), y.to(device)y_hat = net(X)l = loss(y_hat, y)l.backward()optimizer.step()with torch.no_grad():metric.add(l * X.shape[0], d2l.accuracy(y_hat, y), X.shape[0])timer.stop()train_l = metric[0] / metric[2]train_acc = metric[1] / metric[2]if (i + 1) % (num_batches // 5) == 0 or i == num_batches - 1:animator.add(epoch + (i + 1) / num_batches,(train_l, train_acc, None))test_acc = evaluate_accuracy_gpu(net, test_iter)animator.add(epoch + 1, (None, None, test_acc))print(f'loss {train_l:.3f}, train acc {train_acc:.3f}, 'f'test acc {test_acc:.3f}')print(f'{metric[2] * num_epochs / timer.sum():.1f} examples/sec 'f'on {str(device)}')lr, num_epochs = 0.9, 10
train_ch6(net, train_iter, test_iter, num_epochs, lr, d2l.try_gpu())
d2l.plt.show()

1.5 结果展示

2.参考文献

[1]王辉,张帆,刘晓凤,等.基于DarkNet-53和YOLOv3的水果图像识别[J].东北师大学报(自然科学版),2020,52(04):60-65.DOI:10.16163/j.cnki.22-1123/n.2020.04.010.
[2]王治国,曹爽,管海燕,等.基于改进SSD的城市地下排水管道缺陷识别算法[J].测绘工程,2024,33(05):7-13.DOI:10.19349/j.cnki.issn1006-7949.2024.05.002.
[3]杨继雯.基于深度学习的监控视频中人员异常行为识别技术[D].西安工业大学,2024.DOI:10.27391/d.cnki.gxagu.2024.000829.