6 逻辑回归（分类）

视频：https://www.bilibili.com/video/BV1Y7411d7Ys?p=6
博客：

• https://blog.csdn.net/bit452/article/details/109680909

视频中截图

说明：
1、 逻辑斯蒂回归和线性模型的明显区别是在线性模型的后面，添加了激活函数(非线性变换)
2、

• 线性loss： 比较两个数值的差异
• Mini-Batch Loss：比较分布的差异：KL散度，cross-entropy交叉熵

说明：预测与标签越接近，BCE损失越小。

Dataset数据集

MNIST

The database of handwritten digits

• Training set: 60,000 examples,
• Test set: 10,000 examples.
• Classes: 10

  import torchvision
  train_set = torchvision.datasets.MNIST(root='../dataset/mnist', train=True, download=True)
  test_set = torchvision.datasets.MNIST(root='../dataset/mnist', train=False, download=True)

CIFAR-10

32x32小图片

• Training set: 50,000 examples,
• Test set: 10,000 examples.
• Classes: 10

  import torchvision
  train_set = torchvision.datasets.CIFAR10(...)
  test_set = torchvision.datasets.CIFAR10(...)
  

代码说明：

1、视频中代码F.sigmoid(self.linear(x))会引发warning，此处更改为torch.sigmoid(self.linear(x))
2、BCELoss - Binary CrossEntropyLoss
BCELoss 是CrossEntropyLoss的一个特例，只用于二分类问题，而CrossEntropyLoss可以用于二分类，也可以用于多分类。
如果是二分类问题，建议BCELoss
3、损失函数（BCELoss） BCE和CE交叉熵损失函数的区别

数据准备

原来：y_data = torch.Tensor([[2.0], [4.0], [6.0]])
现在：y_data = torch.Tensor([[0], [0], [1]]) 分类使用

逻辑线性单元Logistic Regression Unit

BCELoss交叉熵

可视化

# 可视化

import numpy as np
import matplotlib.pyplot as plt
x = np.linspace(0, 10, 200)
x_t = torch.Tensor(x).view((200, 1))
y_t = model(x_t)  # 使用训练好的模型
y = y_t.data.numpy()
plt.plot(x, y)
plt.plot([0, 10], [0.5, 0.5], c='r')
plt.xlabel('Hours')
plt.ylabel('Probability of Pass')
plt.grid()
plt.show()

代码

'''
Description: 逻辑回归
视频：https://www.bilibili.com/video/BV1Y7411d7Ys?p=6
博客： https://blog.csdn.net/bit452/article/details/109680909
Author: HCQ
Company(School): UCAS
Email: 1756260160@qq.com
Date: 2020-12-05 16:53:13
LastEditTime: 2020-12-05 17:12:56
FilePath: /pytorch/PyTorch深度学习实践/06逻辑回归.py
'''


import torch
# import torch.nn.functional as F

# 1 prepare dataset
x_data = torch.Tensor([[1.0], [2.0], [3.0]])
y_data = torch.Tensor([[0], [0], [1]]) # =============================[0], [0], [1]]=========================

# 2 design model using class
class LogisticRegressionModel(torch.nn.Module):
    def __init__(self):
        super(LogisticRegressionModel, self).__init__()
        self.linear = torch.nn.Linear(1,1)

    def forward(self, x):
        # y_pred = F.sigmoid(self.linear(x))
        y_pred = torch.sigmoid(self.linear(x)) # ========================sigmoid=============================
        return y_pred
model = LogisticRegressionModel()

# 3 construct loss and optimizer
# 默认情况下，loss会基于element平均，如果size_average=False的话，loss会被累加。
criterion = torch.nn.BCELoss(size_average = False)  # ==================BCELoss===========================
optimizer = torch.optim.SGD(model.parameters(), lr = 0.01)

# 4 training cycle forward, backward, update
for epoch in range(1000):
    y_pred = model(x_data)
    loss = criterion(y_pred, y_data)
    print(epoch, loss.item())

    optimizer.zero_grad()
    loss.backward()
    optimizer.step()

print('w = ', model.linear.weight.item())
print('b = ', model.linear.bias.item())

x_test = torch.Tensor([[4.0]])
y_test = model(x_test)
print('y_pred = ', y_test.data)



# 可视化

import numpy as np
import matplotlib.pyplot as plt
x = np.linspace(0, 10, 200)
x_t = torch.Tensor(x).view((200, 1))
y_t = model(x_t)  # 使用训练好的模型
y = y_t.data.numpy()
plt.plot(x, y)
plt.plot([0, 10], [0.5, 0.5], c='r')
plt.xlabel('Hours')
plt.ylabel('Probability of Pass')
plt.grid()
plt.show()