最大池化、非线性激活、线性层

一、最大池化原理

二、最大池化实例

import torch
import torchvision
from torch import nn
from torch.nn import MaxPool2d
from torch.utils.data import DataLoader
from torch.utils.tensorboard import SummaryWriter

dataset = torchvision.datasets.CIFAR10("../chihua",train=False,
                               download=True,transform=torchvision.transforms.ToTensor()) # 对数据集的操作
dataloader = DataLoader(dataset,batch_size=64) # 加载数据集

 构建最大池化神经网络：

class SUN(nn.Module):
    def __init__(self):
        super(SUN, self).__init__()
        self.maxpool1 = MaxPool2d(kernel_size=3, ceil_mode=False)

    def forward(self, input):
        output = self.maxpool1(input)
        return output

sun = SUN()

使用tensorboard显示图片：

writer = SummaryWriter("../logs_maxpool")
step = 0

for data in dataloader:
    imgs,targets = data
    writer.add_image("input", imgs,  step, dataformats="NCHW") # 注意输入的图片，可能出现数据类型的不匹配
    output = sun(imgs)
    writer.add_image("output", output, step, dataformats="NCHW") # 数据通道的设置
    step +=1

writer.close()

显示的结果：

 池化的作用，减小了像素，但是对应的，变得更加模糊。

三、非线性激活

非线性激活的作用，就是主要是给模型加上一些非线性特征，非线性特征越多，才能训练出符合各种特征的模型，提高模型的泛化能力。

四、非线性激活的实例

import torch
import torchvision
from torch import nn
from torch.nn import ReLU, Sigmoid
from torch.utils.data import DataLoader
from torch.utils.tensorboard import SummaryWriter

input = torch.tensor([[1,-0.5],
                      [-1,3]])
input = torch.reshape(input,(-1, 1, 2, 2))
print(input.shape)

dataset = torchvision.datasets.CIFAR10("../datas", train = False, download=True,
                                       transform=torchvision.transforms.ToTensor())
dataloader = DataLoader(dataset, batch_size=64)

class SUN(nn.Module):
    def __init__(self):
        super(SUN, self).__init__()
        self.relu1 = ReLU() # 添加对应的网络
        self.sigmoid = Sigmoid()

    def forward (self, input):
        output = self.sigmoid(input) # 使用了Sigmoid函数
        return output

sun = SUN()
step = 0
write = SummaryWriter("../logs_relu")
for data in dataloader:
    imgs,targets = data
    write.add_image("input", imgs, global_step=step)
    output = sun(imgs)
    write.add_image("output",output,global_step=step)
    step +=1

write.close()

结果

五、 线性层

主要作用是通过线性变换将输入数据映射到一个新的空间，改变数据的维度，便于后续层进一步处理。‌

六、线性层实例

import torch
import torchvision
from torch import nn
from torch.nn import Linear
from torch.utils.data import DataLoader
from torch.utils.tensorboard import SummaryWriter

dataset = torchvision.datasets.CIFAR10("../datalinear", train=False,
                                       transform=torchvision.transforms.ToTensor(), download=True)

dataloader = DataLoader(dataset, batch_size=64, drop_last=True)# 此处特别注意，要设置该参数，否则出现报错：

RuntimeError: mat1 and mat2 shapes cannot be multiplied (1x49152 and 196608x10)

class SUN(nn.Module):
    def __init__(self):
        super(SUN, self).__init__()
        self.linear1 = Linear(196608, 10)

    def forward(self,input):
        output = self.linear1(input)
        return output

sun = SUN()


# writer = SummaryWriter("../logslinear")
# step = 0

for data in dataloader:
    imgs, targets = data
    print(imgs.shape)
    # output = torch.reshape(imgs, (1, 1, 1, -1))
    # 展平
    output = torch.flatten(imgs)
    print(output.reshape)
    output = sun(output)
    print(output.shape)

结果：

 将196608的in_future输出out_future变为10。