Python深度学习pyTorch权重衰减与L2范数正则化解析

下面进行一个高维线性实验

假设我们的真实方程是：

假设feature数200，训练样本和测试样本各20个

模拟数据集

num_train,num_test = 10,10
num_features = 200
true_w = torch.ones((num_features,1),dtype=torch.float32) * 0.01
true_b = torch.tensor(0.5)
samples = torch.normal(0,1,(num_train+num_test,num_features))
noise = torch.normal(0,0.01,(num_train+num_test,1))
labels = samples.matmul(true_w) + true_b + noise
train_samples, train_labels= samples[:num_train],labels[:num_train]
test_samples, test_labels = samples[num_train:],labels[num_train:]

定义带正则项的loss function

def loss_function(predict,label,w,lambd):
   loss = (predict - label) ** 2
   loss = loss.mean() + lambd * (w**2).mean()
   return loss

画图的方法

def semilogy(x_val,y_val,x_label,y_label,x2_val,y2_val,legend):
   plt.figure(figsize=(3,3))
   plt.xlabel(x_label)
   plt.ylabel(y_label)
   plt.semilogy(x_val,y_val)
   if x2_val and y2_val:
       plt.semilogy(x2_val,y2_val)
       plt.legend(legend)
   plt.show()

拟合和画图

def fit_and_plot(train_samples,train_labels,test_samples,test_labels,num_epoch,lambd):
   w = torch.normal(0,1,(train_samples.shape[-1],1),requires_grad=True)
   b = torch.tensor(0.,requires_grad=True)
   optimizer = torch.optim.Adam([w,b],lr=0.05)
   train_loss = []
   test_loss = []
   for epoch in range(num_epoch):
       predict = train_samples.matmul(w) + b
       epoch_train_loss = loss_function(predict,train_labels,w,lambd)
       optimizer.zero_grad()
       epoch_train_loss.backward()
       optimizer.step()
       test_predict = test_sapmles.matmul(w) + b
       epoch_test_loss = loss_function(test_predict,test_labels,w,lambd)
       train_loss.append(epoch_train_loss.item())
       test_loss.append(epoch_test_loss.item())
   semilogy(range(1,num_epoch+1),train_loss,'epoch','loss',range(1,num_epoch+1),test_loss,['train','test'])

可以发现加了正则项的模型，在测试集上的loss确实下降了

以上就是Python深度学习pyTorch权重衰减与L2范数正则化解析的详细内容，更多关于Python pyTorch权重与L2范数正则化的资料请关注脚本之家其它相关文章！

来源：https://blog.csdn.net/qq_43152622/article/details/116937183