（十二）损失函数与反向传播

最新推荐文章于 2024-11-13 20:34:39 发布

千禧霓虹

最新推荐文章于 2024-11-13 20:34:39 发布

阅读量377

点赞数 8

分类专栏： PyTorch快速入门文章标签： pytorch 神经网络深度学习人工智能

本文链接：https://blog.csdn.net/qq_46093326/article/details/135693908

版权

PyTorch快速入门专栏收录该内容

16 篇文章 0 订阅

订阅专栏

本文介绍了常见的三种损失函数（L1Loss、MSELoss和CrossEntropyLoss），并展示了如何在PyTorch中实现它们的代码实例。随后，通过一个简单的神经网络模型和CIFAR10数据集，演示了反向传播和梯度下降的过程。

摘要由CSDN通过智能技术生成

1.常见的损失函数

1.1 L1Loss

在这里插入图片描述

1.2 MSELoss

在这里插入图片描述

1.3 CrossEntropyLoss

在这里插入图片描述

1.4 代码实战

from torch import nn
import torch

input_tensor = torch.tensor([1, 2, 3], dtype=torch.float)
target_tensor = torch.tensor([1, 2, 5], dtype=torch.float)
inputs = torch.reshape(input_tensor, (1, 1, 1, 3))
targets = torch.reshape(target_tensor, (1, 1, 1, 3))

# L1Loss
loss_L1 = nn.L1Loss(reduction='sum')
result = loss_L1(inputs, targets)
print(result)

# MSELoss
loss_MSE = nn.MSELoss()
result_MSE = loss_MSE(inputs, targets)
print(result_MSE)

# CrossEntropyLoss
x = torch.tensor([0.1, 0.2, 0.3])
y = torch.tensor([1])
x = torch.reshape(x, (1, 3))
loss_cross = nn.CrossEntropyLoss()
result_cross = loss_cross(x, y)
print(result_cross)

2.反向传播

2.1 梯度下降

在这里插入图片描述

2.2 代码实战

from torch import nn
from torchvision import datasets, transforms
from torch.utils.data import DataLoader


class My_Module(nn.Module):
    def __init__(self, *args, **kwargs) -> None:
        super().__init__(*args, **kwargs)
        self.seq = nn.Sequential(
            nn.Conv2d(in_channels=3, out_channels=32, kernel_size=5, padding=2),
            nn.MaxPool2d(kernel_size=2),
            nn.Conv2d(in_channels=32, out_channels=32, kernel_size=5, padding=2),
            nn.MaxPool2d(kernel_size=2),
            nn.Conv2d(in_channels=32, out_channels=64, kernel_size=5, padding=2),
            nn.MaxPool2d(kernel_size=2),
            nn.Flatten(),
            nn.Linear(in_features=1024, out_features=64),
            nn.Linear(in_features=64, out_features=10)
        )

    def forward(self, x):
        output = self.seq(x)
        return output


my_module = My_Module()
loss = nn.CrossEntropyLoss()
test_dataset = datasets.CIFAR10(root="datasets", transform=transforms.ToTensor(), download=True)
test_dataloader = DataLoader(dataset=test_dataset, batch_size=1, shuffle=True, drop_last=False)
for data in test_dataloader:
    imgs, targets = data
    outputs = my_module(imgs)
    result_loss = loss(outputs, targets)
    result_loss.backward()