在机器学习中,L1正则化、L2正则化和Elastic Net正则化是用来避免过拟合的技术,它们通过在损失函数中添加一个惩罚项来实现。

正则化介绍

L1 正则化(Lasso回归)

L1 正则化通过向损失函数添加参数的绝对值的和来实施惩罚,公式可以表示为:

其中 L0 是原始的损失函数,λ 是正则化强度,wi是模型参数。

L1 正则化的特点是它可以产生稀疏模型,即许多模型参数会被设置为零。这种特性使得L1正则化不仅可以防止过拟合,还可以进行特征选择。

L2 正则化(Ridge回归)

L2 正则化通过添加参数的平方和来施加惩罚,公式为:

λ 控制着正则化的强度。

L2 正则化倾向于让参数值趋近于零但不会完全为零,这有助于处理参数值过大的问题,从而减少模型在训练数据上的过拟合。

Elastic Net 正则化

Elastic Net 正则化是L1和L2正则化的组合,它在损失函数中同时添加了L1和L2惩罚项,公式为:

这种方法结合了L1和L2的优点,既可以产生稀疏模型,也可以平滑模型参数。

在实际应用中,Elastic Net特别适合于那些特征数量多于样本数量,或者特征之间高度相关的情况。

在sklearn中,我们可以使用内置的回归函数来实现

Lasso回归是应用L1正则化的典型模型。它可以通过

Lasso

类实现;Ridge回归使用L2正则化。它可以通过

Ridge

类来实现;Elastic Net回归结合了L1和L2正则化。它通过

ElasticNet

类实现

Pytorch代码实现

但是这些都是最简单的线性回归的扩展,通过上面的介绍,我们看到这些正则化的方式都是通过修改模型本身的权重来实现的,所以我们可以在MLP上也使用这些正则化的方法,下面我们将使用Pytorch来演示这个步骤

首先我们看下L1

 importos
 importtorch
 fromtorchimportnn
 fromtorchvision.datasetsimportMNIST
 fromtorch.utils.dataimportDataLoader
 fromtorchvisionimporttransforms
 
 classMLP(nn.Module):
   '''
     Multilayer Perceptron.
   '''
   def__init__(self):
     super().__init__()
     self.layers=nn.Sequential(
       nn.Flatten(),
       nn.Linear(28*28*1, 64),
       nn.ReLU(),
       nn.Linear(64, 32),
       nn.ReLU(),
       nn.Linear(32, 10)
     )
 
 
   defforward(self, x):
     '''Forward pass'''
     returnself.layers(x)
   
   defcompute_l1_loss(self, w):
       returntorch.abs(w).sum()
   
   
 if__name__=='__main__':
   
   # Set fixed random number seed
   torch.manual_seed(42)
   
   # Prepare CIFAR-10 dataset
   dataset=MNIST(os.getcwd(), download=True, transform=transforms.ToTensor())
   trainloader=torch.utils.data.DataLoader(dataset, batch_size=10, shuffle=True, num_workers=1)
   
   # Initialize the MLP
   mlp=MLP()
   
   # Define the loss function and optimizer
   loss_function=nn.CrossEntropyLoss()
   optimizer=torch.optim.Adam(mlp.parameters(), lr=1e-4)
   
   # Run the training loop
   forepochinrange(0, 5): # 5 epochs at maximum
     
     # Print epoch
     print(f'Starting epoch {epoch+1}')
     
     # Iterate over the DataLoader for training data
     fori, datainenumerate(trainloader, 0):
       
       # Get inputs
       inputs, targets=data
       
       # Zero the gradients
       optimizer.zero_grad()
       
       # Perform forward pass
       outputs=mlp(inputs)
       
       # Compute loss
       loss=loss_function(outputs, targets)
       
       # Compute L1 loss component
       l1_weight=1.0
       l1_parameters= []
       forparameterinmlp.parameters():
           l1_parameters.append(parameter.view(-1))
       l1=l1_weight*mlp.compute_l1_loss(torch.cat(l1_parameters))
       
       # Add L1 loss component
       loss+=l1
       
       # Perform backward pass
       loss.backward()
       
       # Perform optimization
       optimizer.step()
       
       # Print statistics
       minibatch_loss=loss.item()
       ifi%500==499:
           print('Loss after mini-batch %5d: %.5f (of which %.5f L1 loss)'%
                 (i+1, minibatch_loss, l1))
           current_loss=0.0
 
   # Process is complete.
   print('Training process has finished.')

我们在本身的一个简单的MLP中增加了一个

compute_l1_loss

方法,在我们计算完基本的损失后,还会计算模型参数的L1 损失,然后与基本损失相加,最后使用这个最终损失来进行反向传播。

L2正则化也很容易。我们不取权重值的绝对值,而是取它们的平方。

 importos
 importtorch
 fromtorchimportnn
 fromtorchvision.datasetsimportMNIST
 fromtorch.utils.dataimportDataLoader
 fromtorchvisionimporttransforms
 
 classMLP(nn.Module):
   '''
     Multilayer Perceptron.
   '''
   def__init__(self):
     super().__init__()
     self.layers=nn.Sequential(
       nn.Flatten(),
       nn.Linear(28*28*1, 64),
       nn.ReLU(),
       nn.Linear(64, 32),
       nn.ReLU(),
       nn.Linear(32, 10)
     )
 
 
   defforward(self, x):
     '''Forward pass'''
     returnself.layers(x)
   
   defcompute_l2_loss(self, w):
       returntorch.square(w).sum()
   
   
 if__name__=='__main__':
   
   # Set fixed random number seed
   torch.manual_seed(42)
   
   # Prepare CIFAR-10 dataset
   dataset=MNIST(os.getcwd(), download=True, transform=transforms.ToTensor())
   trainloader=torch.utils.data.DataLoader(dataset, batch_size=10, shuffle=True, num_workers=1)
   
   # Initialize the MLP
   mlp=MLP()
   
   # Define the loss function and optimizer
   loss_function=nn.CrossEntropyLoss()
   optimizer=torch.optim.Adam(mlp.parameters(), lr=1e-4)
   
   # Run the training loop
   forepochinrange(0, 5): # 5 epochs at maximum
     
     # Print epoch
     print(f'Starting epoch {epoch+1}')
     
     # Iterate over the DataLoader for training data
     fori, datainenumerate(trainloader, 0):
       
       # Get inputs
       inputs, targets=data
       
       # Zero the gradients
       optimizer.zero_grad()
       
       # Perform forward pass
       outputs=mlp(inputs)
       
       # Compute loss
       loss=loss_function(outputs, targets)
       
       # Compute l2 loss component
       l2_weight=1.0
       l2_parameters= []
       forparameterinmlp.parameters():
           l2_parameters.append(parameter.view(-1))
       l2=l2_weight*mlp.compute_l2_loss(torch.cat(l2_parameters))
       
       # Add l2 loss component
       loss+=l2
       
       # Perform backward pass
       loss.backward()
       
       # Perform optimization
       optimizer.step()
       
       # Print statistics
       minibatch_loss=loss.item()
       ifi%500==499:
           print('Loss after mini-batch %5d: %.5f (of which %.5f l2 loss)'%
                 (i+1, minibatch_loss, l2))
           current_loss=0.0
 
   # Process is complete.
   print('Training process has finished.')

最终的计算过程和L1正则化一样,只不过是计算附加损失的方法不同。

对于L2的正则化Pytorch的Adam优化器有一个官方的参数,叫做权重衰减 weight_decay

 optimizer = torch.optim.Adam(mlp.parameters(), lr=1e-4, weight_decay=1.0)

你可能不知道他和L2的关系,但是你一定用到过,所以我们这样一解释就非常明白了对吧

最后就是Elastic Net (L1 + L2)

 classMLP(nn.Module):
   '''
     Multilayer Perceptron.
   '''
   def__init__(self):
     super().__init__()
     self.layers=nn.Sequential(
       nn.Flatten(),
       nn.Linear(28*28*1, 64),
       nn.ReLU(),
       nn.Linear(64, 32),
       nn.ReLU(),
       nn.Linear(32, 10)
     )
 
 
   defforward(self, x):
     '''Forward pass'''
     returnself.layers(x)
   
   defcompute_l1_loss(self, w):
       returntorch.abs(w).sum()
   
   defcompute_l2_loss(self, w):
       returntorch.square(w).sum()
   
   
 if__name__=='__main__':
   
   # Set fixed random number seed
   torch.manual_seed(42)
   
   # Prepare CIFAR-10 dataset
   dataset=MNIST(os.getcwd(), download=True, transform=transforms.ToTensor())
   trainloader=torch.utils.data.DataLoader(dataset, batch_size=10, shuffle=True, num_workers=1)
   
   # Initialize the MLP
   mlp=MLP()
   
   # Define the loss function and optimizer
   loss_function=nn.CrossEntropyLoss()
   optimizer=torch.optim.Adam(mlp.parameters(), lr=1e-4)
   
   # Run the training loop
   forepochinrange(0, 5): # 5 epochs at maximum
     
     # Print epoch
     print(f'Starting epoch {epoch+1}')
     
     # Iterate over the DataLoader for training data
     fori, datainenumerate(trainloader, 0):
       
       # Get inputs
       inputs, targets=data
       
       # Zero the gradients
       optimizer.zero_grad()
       
       # Perform forward pass
       outputs=mlp(inputs)
       
       # Compute loss
       loss=loss_function(outputs, targets)
       
       # Specify L1 and L2 weights
       l1_weight=0.3
       l2_weight=0.7
       
       # Compute L1 and L2 loss component
       parameters= []
       forparameterinmlp.parameters():
           parameters.append(parameter.view(-1))
       l1=l1_weight*mlp.compute_l1_loss(torch.cat(parameters))
       l2=l2_weight*mlp.compute_l2_loss(torch.cat(parameters))
       
       # Add L1 and L2 loss components
       loss+=l1
       loss+=l2
       
       # Perform backward pass
       loss.backward()
       
       # Perform optimization
       optimizer.step()
       
       # Print statistics
       minibatch_loss=loss.item()
       ifi%500==499:
           print('Loss after mini-batch %5d: %.5f (of which %.5f L1 loss; %0.5f L2 loss)'%
                 (i+1, minibatch_loss, l1, l2))
           current_loss=0.0
 
   # Process is complete.
   print('Training process has finished.')

也非常的简单,并且我们可以设置两个权重,就是L1和L2的占比,使用不同的加权,可以获得更好的结果。

总结

这篇文章是要是为了介绍L1, L2和Elastic Net (L1+L2)正则化在理论上是如何工作的。并且我们也在PyTorch中使用了L1, L2和Elastic Net (L1+L2)正则化。这三种正则化方法在不同的情况和数据集上有不同的效果,选择哪种正则化方法取决于具体的应用场景和数据特性。

https://avoid.overfit.cn/post/c99ec105e41c4a71a0a1a29735245944


deephub
111 声望88 粉丝