使用pytorch实现SVM

实现SVM的步骤如下：

1. 导入必要的库

import torch
import numpy as np
from sklearn.datasets import make_blobs
from sklearn.preprocessing StandardScaler
from sklearn.model_selection import train_test

2. 生成数据集

X, y = make_blobs(n_samples=100, centers=2, random_state=0)
X = StandardScaler().fit_transform(X)
y = np.where(y == 0, -1, 1)

3. 划分数据集

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=0)

4. 定义SVM模型

class SVM(torch.nn.Module):
    def __init__(self, input_dim):
        super(SVM, self).__init__()
        self.linear = torch.nn.Linear(input_dim, 1)
        
    def forward(self, x):
        return self.linear(x)

5. 初始化模型和优化器

model = SVM(X.shape[1])
optimizer = torch.optim.SGD(model.parameters(), lr=0.01)

6. 训练模型

epochs = 200

for epoch in range(epochs):
    epoch_loss = 0
    for i in range(X_train.shape[0]):
        optimizer.zero_grad()
        xi = torch.FloatTensor(X_train[i])
        yi = torch.FloatTensor([y_train[i]])
        output = yi * model(xi)
        loss = torch.max(torch.zeros(1), 1 - output)
        epoch_loss += loss
        loss.backward()
        optimizer.step()
    print('Epoch {}, Loss {}'.format(epoch, epoch_loss))

7. 测试模型

correct = 0
for i in range(X_test.shape[0]):
    xi = torch.FloatTensor(X_test[i])
    yi = torch.FloatTensor([y_test[i]])
    output = yi * model(xi)
    prediction = torch.sign(output).item()
    if prediction == y_test[i]:
        correct += 1

accuracy = correct / X_test.shape[0]
print('Accuracy:', accuracy)

完整代码如下：

import torch
import numpy as np
from sklearn.datasets import make_blobs
from sklearn.preprocessing import StandardScaler
from sklearn.model_selection import train_test_split

X, y = make_blobs(n_samples=100, centers=2, random_state=0)
X = StandardScaler().fit_transform(X)
y = np.where(y == 0, -1, 1)

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=0)

class SVM(torch.nn.Module):
    def __init__(self, input_dim):
        super(SVM, self).__init__()
        self.linear = torch.nn.Linear(input_dim, 1)
        
    def forward(self, x):
        return self.linear(x)

model = SVM(X.shape[1])
optimizer = torch.optim.SGD(model.parameters(), lr=0.01)

epochs = 200

for epoch in range(epochs):
    epoch_loss = 0
    for i in range(X_train.shape[0]):
        optimizer.zero_grad()
        xi = torch.FloatTensor(X_train[i])
        yi = torch.FloatTensor([y_train[i]])
        output = yi * model(xi)
        loss = torch.max(torch.zeros(1), 1 - output)
        epoch_loss += loss
        loss.backward()
        optimizer.step()
    print('Epoch {}, Loss {}'.format(epoch, epoch_loss))

correct = 0
for i in range(X_test.shape[0]):
    xi = torch.FloatTensor(X_test[i])
    yi = torch.FloatTensor([y_test[i]])
    output = yi * model(xi)
    prediction = torch.sign(output).item()
    if prediction == y_test[i]:
        correct += 1

accuracy = correct / X_test.shape[0]
print('Accuracy:', accuracy)