可以进行降维也可以进行升维,其实就是矩阵的计算。
升维会出现一个问题,就是超参数的查找和应用。
隐层越多,学习能力越强,单数学习能力太强也不是很好,会把模型里面的噪音也学习进来。
(比如:学习能力太强了,就类似于把书本背死了)
数据集可以在自行下载:链接: https://pan.baidu.com/s/1Ku5c99yDHNFMt8EJAcF5LA 提取码: n4xh
如果失效了可以私信我
torch.sigmoid、torch.nn.Sigmoid和torch.nn.functional.sigmoid的区别
import torch
import numpy as np
import matplotlib.pyplot as plt
import torch.nn.functional as F
xy = np.loadtxt('diabetes.csv.gz',delimiter=',',dtype=np.float32) # delimiter分割 dtype=np.float32一般都使用32位的浮点数
x_data = torch.from_numpy(xy[:,:-1])
y_data = torch.from_numpy(xy[:,[-1]]) # [-1]拿出来的是一个矩阵而不是向量,要保证在计算的时候是矩阵
# print(x_data,y_data)
class Model(torch.nn.Module):
def __init__(self):
super(Model,self).__init__()
self.linear1 = torch.nn.Linear(8,7)
self.linear2 = torch.nn.Linear(7,6)
self.linear3 = torch.nn.Linear(6,5)
self.linear4 = torch.nn.Linear(5,4)
self.linear5 = torch.nn.Linear(4,3)
self.linear6 = torch.nn.Linear(3,2)
self.linear7 = torch.nn.Linear(2,1)
self.sigmoid = torch.nn.Sigmoid() # nn下的sigmoid pytorch激活函数很多,可以都试试,例如:rule...torch.nn.Rule() //取值0-1
# self.activate = torch.nn.Sigmoid()
def forward(self,x):
# x = self.sigmoid(self.linear1(x))
# x = self.sigmoid(self.linear2(x))
# x = self.sigmoid(self.linear3(x))
x = self.sigmoid(self.linear1(x))
x = self.sigmoid(self.linear2(x))
x = self.sigmoid(self.linear3(x))
x = self.sigmoid(self.linear4(x))
x = self.sigmoid(self.linear5(x))
x = self.sigmoid(self.linear6(x))
x = self.sigmoid(self.linear7(x))
return x
model = Model()
# 损失函数和优化器
criterion = torch.nn.BCELoss(reduction='sum')
optimizer = torch.optim.SGD(model.parameters(),lr = 0.01)
epoch_l = []
loss_l = []
for epoch in range(800):
# forward
y_pred = model(x_data)
loss= criterion(y_pred,y_data)
print(epoch,loss.item())
# backward
optimizer.zero_grad()
loss.backward()
# update
optimizer.step()
epoch_l.append(epoch)
loss_l.append(loss.item())
plt.plot(epoch_l,loss_l,c='r')
plt.show()
比较变态的写了7层,哈哈
可以看出来已经收敛了
关于训练的轮数我也不知道怎么选,有没有朋友可以交流一下经验