深度学习-线性回归模型解析

import torch import matplotlib.pyplot as plt #画图的 import random #随机 def create_data(w,b,data_num) : #生成数据 x = torch.normal(0,1,(data_num,len(w))) y = torch.matmul(x,w) +b #matmul表示矩阵相乘 noise = torch.normal(0,0.01,y.shape) #噪声要加到y上 y+=noise return x,y num = 500 true_w = torch.tensor([8.1,2,2,4]) true_b = torch.tensor(1.1) X,Y = create_data(true_w,true_b,num)plt.scatter(X[:,2],Y,1) plt.show() def data_provider(data,label,batchsize): #每次访问这个函数，就能提供一批数据 length = len(label) indices = list(range(length)) #变为0-500的列表 #不按顺序取数，把数据打乱 random.shuffle(indices) for each in range(0,length,batchsize): get_indices = indices[each:each+batchsize] get_data = data[get_indices] get_label = label[get_indices] yield get_data,get_label #有存档点的returnbatchsize = 16 for batch_x,batch_y in data_provider(X,Y,batchsize): print(batch_x,batch_y) # break def fun(x,w,b): pred_y = torch.matmul(x,w) +b return pred_y def maeLoss(pre_y,y): return torch.sum(abs(pre_y-y))/len(y)def sgd(paras,lr): #随机梯度下降。更新参数 with torch.no_grad(): #属于这句代码的部分，不计算梯度 for para in paras: para -=para.grad * lr para.grad.zero_() #使用过的梯度，归0lr = 0.03 #学习率 w_0 = torch.normal(0,0.01,true_w.shape,requires_grad=True) #这个w需要计算梯度 b_0 = torch.tensor(0.01,requires_grad=True) print(w_0,b_0) epochs = 50 for epoch in range(epochs): data_loss = 0 for batch_x,batch_y in data_provider(X,Y,batchsize): pred_y = fun(batch_x,w_0,b_0) loss = maeLoss(pred_y,batch_y) loss.backward() sgd([w_0,b_0],lr) data_loss +=loss print("epoch %03d: loss: %.6f"%(epoch,data_loss)) print("真实的函数值是：",true_w,true_b) print("训练得到的参数值是：",w_0,b_0) idx = 2 plt.plot(X[:,idx].detach().numpy(),X[:,idx].detach().numpy()*w_0[idx].detach().numpy()+b_0.detach().numpy()) plt.scatter(X[:,idx],Y,1) plt.show()