pytorch: 学习笔记4, pytorch实现线性回归

pytorch实现线性回归

线性回归应用：如计算房价与面积/房龄的线性函数关系，参数面积越大，一般房价越高(正比，y=kx正斜率k)；年代越长，一般房价越低(反比，y=kx负斜率k)。假设房价与两个参数呈线性关系，则初始化参数时，取 true_w = [2, -3.4]

import numpy as np
import torch
import random
from matplotlib import pyplot as plt


def show(sample, labels):
    print('show')
    plt.scatter(sample, labels, 1)
    plt.show()

def data_iter(batch_size, features, labels):
    num_examples = len(features)
    indices = list(range(num_examples))
    random.shuffle(indices)  # 样本的读取顺序是随机的
    for i in range(0, num_examples, batch_size):
        j = torch.LongTensor(indices[i: min(i + batch_size, num_examples)]) # 最后一次可能不足一个batch
        yield  features.index_select(0, j), labels.index_select(0, j)

def linreg(X, w, b):
    return torch.mm(X, w) + b

def squared_loss(y_hat, y):
    # 注意这里返回的是向量, 另外, pytorch里的MSELoss并没有除以 2
    return (y_hat - y.view(y_hat.size())) ** 2 / 2

def sgd(params, lr, batch_size):
    for param in params:
        param.data -= lr * param.grad / batch_size # 注意这里更改param时用的param.data


def main():
    # 生成 1000*2 数据
    num_inputs = 2
    num_examples = 1000
    true_w = [2, -3.4]
    true_b = 4.2
    features = torch.randn(num_examples, num_inputs, dtype=torch.float32)
    labels = true_w[0] * features[:, 0] + true_w[1] * features[:, 1] + true_b
    labels += torch.tensor(np.random.normal(0, 0.01, size=labels.size()), dtype=torch.float32)
    # show(features[:, 0].numpy(), labels.numpy())
    # show(features[:, 1].numpy(), labels.numpy())

    # 按照batch_size取数据
    batch_size = 10
    # for X, y in data_iter(batch_size, features, labels):
    #     print(X, y)
    #     break

    # 随机初始化w b
    w = torch.tensor(np.random.normal(0, 0.01, (num_inputs, 1)), dtype=torch.float32)
    b = torch.zeros(1, dtype=torch.float32)
    w.requires_grad_(requires_grad=True)
    b.requires_grad_(requires_grad=True)
    print("w b : ", '\n', w, '\n', b)

    # 设置超参数 网络与损失函数
    lr = 0.03
    num_epochs = 3
    net = linreg
    loss = squared_loss

    # 训练
    for epoch in range(num_epochs):  # 训练模型一共需要num_epochs个迭代周期
        # 在每一个迭代周期中，会使用训练数据集中所有样本一次（假设样本数能够被批量大小整除）。X
        # 和y分别是小批量样本的特征和标签
        for X, y in data_iter(batch_size, features, labels):
            l = loss(net(X, w, b), y).sum()  # l是有关小批量X和y的损失
            l.backward()  # 小批量的损失对模型参数求梯度
            sgd([w, b], lr, batch_size)  # 使用小批量随机梯度下降迭代模型参数，更新w, b

            # 不要忘了梯度清零
            w.grad.data.zero_()
            b.grad.data.zero_()
        train_l = loss(net(features, w, b), labels)
        print('epoch %d, loss %f' % (epoch + 1, train_l.mean().item()))  # .item()得到一个元素张量里面的元素值

    print(true_w, '\n', w)
    print(true_b, '\n', b)


if __name__ == '__main__':
    main()

结果：

w b :  
 tensor([[ 0.0056],
        [-0.0033]], requires_grad=True) 
 tensor([0.], requires_grad=True)
epoch 1, loss 0.034380
epoch 2, loss 0.000122
epoch 3, loss 0.000049
[2, -3.4] 
 tensor([[ 1.9996],
        [-3.3998]], requires_grad=True)
4.2 
 tensor([4.2002], requires_grad=True)

参考学习，把学习中的知识整合，并非自己实现。
参考：https://tangshusen.me/Dive-into-DL-PyTorch/#/chapter03_DL-basics/3.2_linear-regression-scratch