卷积网络与全连接网络比较分析

from torchvision import datasetsfrom torchvision.transforms import ToTensorfrom torch import nnimport torchfrom torch.utils.data import DataLoaderimport matplotlib.pyplot as pltfrom collections import defaultdictclass MyNet(nn.Module): def __init__(self) -> None: super().__init__() self.flatten = nn.Flatten() self.fc1 = nn.Linear(in_features=784, out_features=512) self.fc2 = nn.Linear(in_features=512, out_features=10) def forward(self, x): x = self.flatten(x) x = self.fc1(x) out = self.fc2(x) return outdef train(dataloader,net,loss_fn,optimizer): size = len((dataloader.dataset)) epoch_loss=0.0 batch_num=len(dataloader) net.train() correct = 0 for batch_ind, (X,y) in enumerate(dataloader): X,y=X.to(device),y.to(device) pred = net(X) loss=loss_fn(pred,y) optimizer.zero_grad() loss.backward() optimizer.step() epoch_loss+=loss.item() correct+=(pred.argmax(1)==y).type(torch.float).sum().item() if batch_ind % 100==0: print(f'[{batch_ind+1 :>5d} / {batch_num :>5d}], loss:{loss.item()}') avg_loss=epoch_loss/batch_num avg_accurcy = correct / size return avg_accurcy,avg_lossdef test(dataloder,net,loss): size=len(dataloder.dataset) batch_num = len(dataloder) net.eval() losses=0 correct = 0 with torch.no_grad(): for X,y in dataloder: X,y = X.to(device),y.to(device) pred = net(X) loss = loss_fn(pred,y) losses+=loss.item() correct+=(pred.argmax(1) ==y).type(torch.int).sum().item() accuracy = correct / size avg_loss = losses /batch_num print(f'accuracy is {accuracy*100}%') return accuracy,avg_lossclass getData: def __init__(self): self.train_ds = datasets.MNIST( root='data', download=True, train=True, transform=ToTensor(), ) self.text_ds = datasets.MNIST( root='data', download=True, train=False, transform=ToTensor(), ) self.train_loader = DataLoader( dataset=self.train_ds, batch_size=128, shuffle=True, ) self.test_loader = DataLoader( dataset=self.text_ds, batch_size=128, )if __name__=='__main__': batch_size=128 device='cuda' if torch.cuda.is_available() else 'cpu' train_ds = datasets.MNIST( root='data', download=True, train=True, transform=ToTensor(), ) text_ds = datasets.MNIST( root='data', download=True, train=False, transform=ToTensor(), ) train_ds,val_ds= torch.utils.data.random_split(train_ds,[50000,10000]) # (3) train_loader = DataLoader( dataset=train_ds, batch_size=batch_size, shuffle=True, ) val_loder = DataLoader( dataset=val_ds, batch_size=batch_size, ) test_loader = DataLoader( dataset=text_ds, batch_size=batch_size, ) net=MyNet().to(device) optimizer=torch.optim.SGD(net.parameters(),lr=0.15) loss_fn=nn.CrossEntropyLoss() train_loss=[] train_acc_list,train_loss_list,val_acc_list,val_loss_list=[],[],[],[] best_acc=0 for epoch in range(10): print('-'*50) print(f'eopch:{epoch+1}') train_accuracy,train_loss=train(train_loader,net,loss_fn,optimizer) val_accuracy,val_loss =test(train_loader,net,loss_fn) print(f'train acc:{train_accuracy},train_val:{train_loss}') train_acc_list.append(train_accuracy) train_loss_list.append(train_loss) val_acc_list.append(val_accuracy) val_loss_list.append(val_loss) if val_accuracy > best_acc: best_acc = val_accuracy torch.save(net.state_dict(),'model_best.pth') net.load_state_dict(torch.load('model_best.pth')) print('the best val_acc is:') test(test_loader,net,loss_fn)

import torchfrom torch import nnx = torch.rand(128,3,28,28)conv1 = nn.Conv2d(in_channels = 3,out_channels = 16,kernel_size = 3,stride = 1,padding = 1)conv2 = nn.Conv2d(in_channels = 16,out_channels = 32,kernel_size = 3,stride = 1,padding = 1)x = conv1(x)out = conv2(x)print(out.shape)