本文共 2924 字，大约阅读时间需要 9 分钟。

PyTorch 神经网络搭建实例

PyTorch 神经网络的搭建通常分为以下几个关键步骤：数据准备、网络架构设计、损失函数与优化器选择以及反向传播与更新。

1. 分类任务示例

import torch
import torch.nn.functional as F
import matplotlib.pyplot as plt
# 数据准备
n_data = 200
x0 = torch.normal(2 * n_data, 1, (100, 2))
y0 = torch.zeros(100)
x1 = torch.normal(-2 * n_data, 1, (100, 2))
y1 = torch.ones(100)
x = torch.cat((x0, x1), 0).type(torch.FloatTensor)
y = torch.cat((y0, y1), 0).type(torch.LongTensor)
# 网络架构定义
class Net(torch.nn.Module):
    def __init__(self, n_feature, n_hidden, n_output):
        super(Net, self).__init__()
        self.hidden = torch.nn.Linear(n_feature, n_hidden)
        self.out = torch.nn.Linear(n_hidden, n_output)
    
    def forward(self, x):
        x = F.relu(self.hidden(x))
        x = self.out(x)
        return x
net = Net(n_feature=2, n_hidden=10, n_output=2)
print(net)
# 优化器与损失函数
optimizer = torch.optim.SGD(net.parameters(), lr=0.02)
loss_func = torch.nn.CrossEntropyLoss()
# 训练过程
plt.ion()
for t in range(100):
    out = net(x)
    loss = loss_func(out, y)
    optimizer.zero_grad()
    loss.backward()
    optimizer.step()
    
    if t % 2 == 0:
        plt.cla()
        prediction = torch.max(out, 1)[1]
        accuracy = (prediction.data.numpy() == y.data.numpy()).sum() / y.size()
        plt.scatter(x.data.numpy()[:, 0], x.data.numpy()[:, 1], c=prediction.data.numpy(), s=100, lw=0, cmap='RdYlGn')
        plt.text(1.5, -4, 'Accuracy: %.2f' % accuracy, fontdict={'size': 20, 'color': 'red'})
        plt.pause(0.1)
plt.ioff()
plt.show()

2. 回归任务示例

import torch
import torch.nn.functional as F
import matplotlib.pyplot as plt
# 数据准备
x_train = torch.unsqueeze(torch.linspace(-1, 1, 100), dim=1).type(torch.FloatTensor)
y_train = x_train.pow(2) + 0.2 * torch.rand(x_train.size())
# 网络架构定义
class Net(torch.nn.Module):
    def __init__(self, n_feature, n_hidden, n_output):
        super(Net, self).__init__()
        self.hidden = torch.nn.Linear(n_feature, n_hidden)
        self.predict = torch.nn.Linear(n_hidden, n_output)
    
    def forward(self, x):
        x = F.relu(self.hidden(x))
        x = self.predict(x)
        return x
net = Net(n_feature=1, n_hidden=10, n_output=1)
print(net)
# 优化器与损失函数
optimizer = torch.optim.SGD(net.parameters(), lr=0.2)
criterion = torch.nn.MSELoss()
# 训练过程
plt.ion()
for t in range(200):
    prediction = net(x_train)
    loss = criterion(prediction, y_train)
    optimizer.zero_grad()
    loss.backward()
    optimizer.step()
    
    if t % 5 == 0:
        plt.cla()
        plt.scatter(x_train.numpy(), y_train.numpy())
        plt.plot(x_train.numpy(), prediction.data.numpy(), 'r-', lw=5)
        plt.text(0.5, 0, 'Loss: %.4f' % loss.item(), fontdict={'size': 20, 'color': 'red'})
        plt.pause(0.1)
plt.ioff()
plt.show()

3. 优化后的代码解释

在上述代码中，主要进行了以下改进：

通过以上代码示例，可以清晰地看到从数据准备到模型训练再到结果可视化的完整流程。每一步都经过精心设计，确保代码既简洁又高效。

转载地址：http://jfxp.baihongyu.com/