58 lines
1.5 KiB
Python
58 lines
1.5 KiB
Python
|
#多项式拟合
|
|||
|
|
import torch
|
|||
|
|
import numpy as np
|
|||
|
|
|
|||
|
|
from FC_ML_Data.FC_ML_Data_Output.Data_Output_File import tensor_to_excel
|
|||
|
|
|
|||
|
|
# 真实多项式系数
|
|||
|
|
true_w = torch.tensor([0.5, 3.0, 2.4]) # 对应x, x², x³项
|
|||
|
|
true_b = 0.9
|
|||
|
|
|
|||
|
|
# 生成训练数据
|
|||
|
|
def make_features(x):
|
|||
|
|
return torch.stack([x**i for i in range(1,4)], dim=1) # 构建x, x², x³特征矩阵
|
|||
|
|
|
|||
|
|
x = torch.linspace(-3, 3, 100)
|
|||
|
|
X = make_features(x)
|
|||
|
|
y = X @ true_w + true_b + torch.randn(x.size()) * 0.5 # 添加噪声
|
|||
|
|
print(x,X,y)
|
|||
|
|
# tensor_to_excel(torch.cat([x, y], dim=-1),"./")
|
|||
|
|
|
|||
|
|
|
|||
|
|
class PolyModel(torch.nn.Module):
|
|||
|
|
def __init__(self):
|
|||
|
|
super().__init__()
|
|||
|
|
self.linear = torch.nn.Linear(3, 1) # 输入3维(x,x²,x³)，输出1维
|
|||
|
|
|
|||
|
|
def forward(self, x):
|
|||
|
|
return self.linear(x)
|
|||
|
|
|
|||
|
|
|
|||
|
|
model = PolyModel()
|
|||
|
|
criterion = torch.nn.MSELoss()
|
|||
|
|
optimizer = torch.optim.SGD(model.parameters(), lr=1e-3)
|
|||
|
|
|
|||
|
|
for epoch in range(1000):
|
|||
|
|
pred = model(X)
|
|||
|
|
loss = criterion(pred.squeeze(), y)
|
|||
|
|
|
|||
|
|
optimizer.zero_grad()
|
|||
|
|
loss.backward()
|
|||
|
|
optimizer.step()
|
|||
|
|
|
|||
|
|
if epoch % 100 == 0:
|
|||
|
|
print(f'Epoch {epoch}, Loss: {loss.item():.4f}')
|
|||
|
|
|
|||
|
|
# 获取训练后的参数
|
|||
|
|
w_pred = model.linear.weight.detach().numpy().flatten()
|
|||
|
|
b_pred = model.linear.bias.detach().numpy()
|
|||
|
|
|
|||
|
|
# print(f"真实参数: w={true_w.numpy()}, b={true_b}")
|
|||
|
|
# print(f"预测参数: w={w_pred}, b={b_pred:.2f}")
|
|||
|
|
|
|||
|
|
# 可视化
|
|||
|
|
import matplotlib.pyplot as plt
|
|||
|
|
plt.scatter(x, y, label='ori')
|
|||
|
|
plt.plot(x, model(X).detach().numpy(), 'r', label='fit')
|
|||
|
|
plt.legend()
|
|||
|
|
plt.show()
|