感知机

感知机是由输入input经过线性变化得到特定输出output(0或1)的结构
多层感知机引入非线性层/隐藏层（激活函数）

下面是多层感知机手动实现：

import torch
from torch import nn
from d2l import torch as d2l


batch_size = 256
#获取数据集
train_iter,test_iter = d2l.load_data_fashion_mnist(batch_size)

num_inputs = 784  #28*28
num_outputs = 10  #10个类别
num_hiddens = 256  #隐藏层神经元个数(超参数)

# 训练层参数
W1 = nn.Parameter(
    torch.randn(num_inputs, num_hiddens, requires_grad=True)#训练时反向传播需要梯度
)
b1 = nn.Parameter(
    torch.zeros(num_hiddens, requires_grad=True)
)

W2 = nn.Parameter(
    torch.randn(num_hiddens, num_outputs, requires_grad=True)
)
b2 = nn.Parameter(
    torch.zeros(num_outputs, requires_grad=True)
)

params = [W1, b1, W2, b2]

#实现激活函数
def relu(X):
    a = torch.zeros_like(X)
    return torch.max(X, a)

#实现模型
def net(X):
    X = X.reshape((-1, num_inputs))  #将输入展平
    H = relu(X @ W1 + b1)  #隐藏层，@表示矩阵乘法
    return H @ W2 + b2  #输出层

num_epochs, lr = 10, 0.1  #训练轮数，学习率
updater = torch.optim.SGD(params, lr=lr)  #优化器
d2l.train_ch3(
    net, train_iter, test_iter, loss=nn.CrossEntropyLoss(),num_epochs=num_epochs,
    updater=updater
)