神经网络梯度下降解决异或问题

 通过神经网络，可以解决异或问题（证明神经网络可以解决线性不可分问题）

import torch
from torch import nn, optim
from torch.nn import functional as F


class NLP(nn.Module):
    def __init__(self, input_dim, hidden_dim, num_class):
        super(NLP, self).__init__()
        self.linear1 = nn.Linear(input_dim, hidden_dim)
        self.activate = F.relu
        self.linear2 = nn.Linear(hidden_dim, num_class)

    def forward(self, inputs):
        hidden = self.linear1(inputs)
        activation = self.activate(hidden)
        outputs = self.linear2(activation)
        log_probs = F.softmax(outputs, dim=1)
        return log_probs


x_train = torch.tensor([[0.0, 0.0], [0.0, 1.0], [1.0, 0.0], [1.0, 1.0]])
y_train = torch.tensor([0, 1, 1, 0])

model = NLP(input_dim=2, hidden_dim=5, num_class=2)

criterion = nn.NLLLoss
criteria = nn.CrossEntropyLoss()
optimizer = optim.SGD(model.parameters(), lr=0.05)  # 学习率为0.05

for epoch in range(10000):      # 训练次数
    y_pred = model(x_train)     # 训练
    loss = criteria(y_pred, y_train)    # 计算损失
    optimizer.zero_grad()   # 将梯度置为0，防止累加
    loss.backward()     # 反向传播
    optimizer.step()    # 更新参数

print("Parameters:")
for name, param in model.named_parameters():    # 参数 值
    print(name, param.data)

y_pred = model(x_train)
print(y_pred)
print("Predicted results:", y_pred.argmax(dim=1))