0.1 学习视频源于:b站:刘二大人《PyTorch深度学习实践》
0.2 本章内容为自主学习总结内容,若有错误欢迎指正!
代码(类比线性回归):
python
# 调用库
import torch
import torch.nn.functional as F
# 数据准备
x_data = torch.Tensor([[1.0], [2.0], [3.0]]) # 训练集输入值
y_data = torch.Tensor([[0], [0], [1]]) # 训练集输出值
# 定义逻辑回归模型
class LogisticRegressionModel(torch.nn.Module):
def __init__(self):
super(LogisticRegressionModel, self).__init__() # 调用父类构造函数
self.linear = torch.nn.Linear(1, 1) # 实例化torch库nn模块的Linear类,特征一维,输出一维
def forward(self, x):
"""
前馈运算
:param x: 输入值
:return: 线性回归预测结果
"""
y_pred = F.sigmoid(self.linear(x))
return y_pred
model = LogisticRegressionModel() # 实例化
criterion = torch.nn.BCELoss(size_average=False) # 损失函数
optimizer = torch.optim.SGD(model.parameters(), lr=0.01) # 优化器------梯度下降SGD
# 训练过程
for epoch in range(1000): # epoch:训练轮次
y_pred = model(x_data)
loss = criterion(y_pred, y_data)
print(epoch, loss.item())
optimizer.zero_grad() # 梯度归零
loss.backward() # 反向传播
optimizer.step() # 权重自动更新
print("w = ", model.linear.weight.item())
print("b = ", model.linear.bias.item())
# 预测过程
x_test = torch.Tensor([[3.5]])
y_test = model(x_test)
print("y_pred = ", y_test.data)
结果:
注:输出结果为类别是1的概率。