pytorch進行mnist識別實戰

mnist實戰

開始使用簡單的全連接層進行mnist手寫數字的識別，識別率最高能到95%，而使用兩層卷積后再全連接，識別率能達到99%

全連接：

import torch
from torch import nn
from torch.nn import functional as F
from torch import optim
import torchvision
from    matplotlib import pyplot as plt
from torch.optim.lr_scheduler import StepLR

#step 1:load dataset

def plot_image(img, label, name):
    fig = plt.figure()
    for i in range(6):
        plt.subplot(2, 3, i + 1)
        plt.tight_layout()
        plt.imshow(img[i][0] * 0.3081 + 0.1307, cmap='gray', interpolation='none')
        plt.title("{}: {}".format(name, label[i].item()))
        plt.xticks([])
        plt.yticks([])
    plt.show()

def plot_curve(data):
    fig = plt.figure()
    plt.plot(range(len(data)), data, color='blue')
    plt.legend(['value'], loc='upper right')
    plt.xlabel('step')
    plt.ylabel('value')
    plt.show()


batch_size=512

train_loader = torch.utils.data.DataLoader(
    torchvision.datasets.MNIST('mnist_data',train=True,download=True,
                               transform=torchvision.transforms.Compose(
                                   [
                                       torchvision.transforms.ToTensor(),
                                       torchvision.transforms.Normalize((0.1307,),(0.3081,))#這里的兩個數字分別是數據集的均值是0.1307，標准差是0.3081
                                   ]
                               )
                               ),
    batch_size=batch_size,shuffle=True
)

test_loader = torch.utils.data.DataLoader(
    torchvision.datasets.MNIST('mnist_data/',train=False,download=True,#是驗證集所以train=False
                               transform=torchvision.transforms.Compose(
                                   [
                                       torchvision.transforms.ToTensor(),
                                       torchvision.transforms.Normalize((0.1307,),(0.3081,))
                                   ]
                               )
                               ),
    batch_size=batch_size,shuffle=False#是驗證集所以無需打亂，shuffle=False
)

# x,y = next(iter(train_loader))
# plot_image(x,y,'example')


#step2: create network


class Net(nn.Module):
    def __init__(self):
        super(Net, self).__init__()

        #wx+b
        self.fc1 = nn.Linear(28*28,256)#256是自己根據經驗隨機設定的
        self.fc2 = nn.Linear(256,64)
        self.fc3 = nn.Linear(64,10)#注意這里的10是最后識別的類別數（最后一層的輸出往往是識別的類別數）

    def forward(self, x):
        #x : [ b 1 28 28]有batch_size張圖片，通道是1維灰度圖像 圖片大小是28*28

        #h1=relu(wx+b)
        x = F.relu(self.fc1(x))#使用relu非線性激活函數包裹
        x = F.relu(self.fc2(x))
        x = F.softmax(self.fc3(x))#由於是多類別識別，所以使用softmax函數
        #x = self.fc3(x)
        return x

net = Net()
optimizer = optim.Adam(net.parameters())
train_loss = []




for epoch in range(5):

    for batch_idx,(x,y) in enumerate(train_loader):#enumerate表示在數據前面加上序號組成元組，默認序號從0開始

        # x :[512 1 28 28]   y : [512]

        #由於這里的x維度為[512 1 28 28]，但是在網絡中第一層就是一個全連接層，維度只能是[b,feature(784)]，所以要把x打平
        #將前面多維度的tensor展平成一維

        # 卷積或者池化之后的tensor的維度為(batchsize，channels，x，y)，其中x.size(0)
        # 指batchsize的值，最后通過x.view(x.size(0), -1)
        # 將tensor的結構轉換為了(batchsize, channels * x * y)，即將（channels，x，y）拉直，然后就可以和fc層連接了

        x = x.view(x.size(0),28*28)
        #輸出之后的維度變為[512,10]
        out=net(x)
        #使用交叉熵損失
        loss = F.cross_entropy(out,y)

        #清零梯度——計算梯度——更新梯度

        #要進行梯度的清零
        optimizer.zero_grad()

        loss.backward()
        #功能是: w` = w-lr*grad
        optimizer.step()

        train_loss.append(loss.item())#將loss保存在trainloss中，而loss.item()表示將tensor 的類型轉換為數值類型

        #打印loss
        if batch_idx % 10 == 0:
            print(epoch,batch_idx,loss.item())


plot_curve(train_loss)

total_correct = 0
for x, y in test_loader:
    x = x.view(x.size(0),28*28)
    out = net(x)
    #out :[512,10]
    pred = out.argmax(dim = 1)
    correct = pred.eq(y).sum().float().item()#當前批次識別對的個數
    total_correct+= correct

total_number = len(test_loader.dataset)
acc = total_correct / total_number
print('test acc',acc)


x,y = next(iter(test_loader))
out = net(x.view(x.size(0),28*28))
pred = out.argmax(dim=1)
plot_image(x,pred,'test')

#optimizer = optim.SGD(net.parameters(),lr=0.1,momentum=0.9)
#test acc 0.8783

#optimizer = optim.Adam(net.parameters())
#test acc 0.9574

加入卷積：

import torch
import argparse
import torch.nn as nn
import matplotlib.pyplot as plt
import torch.optim as optim
import torch.nn.functional as F
from torchvision import datasets,transforms
from torch.optim.lr_scheduler import StepLR

class Net(nn.Module):
    def __init__(self):
        super(Net,self).__init__()
        self.conv1 = nn.Conv2d(1,32,3,1)
        self.conv2 = nn.Conv2d(32,64,3,1)
        self.dropout1 = nn.Dropout2d(0.25)
        self.dropout2 = nn.Dropout2d(0.5)
        self.fc1 = nn.Linear(9216, 128)
        self.fc2 = nn.Linear(128, 10)

    def forward(self,x):
        x = self.conv1(x)
        x = F.relu(x)
        x = self.conv2(x)
        x = F.relu(x)
        #print(x.shape)
        x = F.max_pool2d(x, 2)
        x = self.dropout1(x)
        #print(x.shape)
        x = torch.flatten(x,1)
        #print(x.shape)
        x = self.fc1(x)
        x = F.relu(x)
        x = self.dropout2(x)
        x = self.fc2(x)
        output = F.softmax(x)
        return output

#用來查看經過conv之后進入全連接層的維度
# def main():
#     net = Net()
#
#     tmp = torch.rand(10,1,28,28)
#     out = net.forward(tmp)
#
#
# if __name__=='__main__':
#     main()
# torch.Size([10, 64, 24, 24])
# torch.Size([10, 64, 12, 12])
# torch.Size([10, 9216])

def plot_image(img, label, name):
    fig = plt.figure()
    for i in range(6):
        plt.subplot(2, 3, i + 1)
        plt.tight_layout()
        plt.imshow(img[i][0] * 0.3081 + 0.1307, cmap='gray', interpolation='none')
        plt.title("{}: {}".format(name, label[i].item()))
        plt.xticks([])
        plt.yticks([])
    plt.show()

def train(args,model,device,train_loader,optimizer,epoch):
    model.train()#進入訓練模式來激活dropout層、正則化等的使用
    for batch_idx,(data,target) in enumerate(train_loader):
        data,target = data.to(device),target.to(device)
        optimizer.zero_grad()
        output = model(data)
        loss = F.cross_entropy(output,target)
        loss.backward()
        optimizer.step()
        if batch_idx % args.log_interval ==0:
            print('train Epoch: {} [{}/{} ({:.0f}%)]\tLoss: {:.6f}'.format(
                epoch, batch_idx * len(data), len(train_loader.dataset),
                100. * batch_idx / len(train_loader), loss.item()))
            if args.dry_run:
                break

def test(model, device, test_loader):
    model.eval()
    test_loss = 0
    correct = 0
    with torch.no_grad():
        for data, target in test_loader:
            data, target = data.to(device), target.to(device)
            output = model(data)
            test_loss += F.cross_entropy(output, target, reduction='sum').item()  # sum up batch loss
            pred = output.argmax(dim=1, keepdim=True)  # get the index of the max log-probability
            correct += pred.eq(target.view_as(pred)).sum().item()

    test_loss /= len(test_loader.dataset)

    print('\nTest set: Average loss: {:.4f}, Accuracy: {}/{} ({:.0f}%)\n'.format(
        test_loss, correct, len(test_loader.dataset),
        100. * correct / len(test_loader.dataset)))


def main():
    # Training settings
    parser = argparse.ArgumentParser(description='PyTorch MNIST Example')
    parser.add_argument('--batch-size', type=int, default=64, metavar='N',
                        help='input batch size for training (default: 64)')
    parser.add_argument('--test-batch-size', type=int, default=1000, metavar='N',
                        help='input batch size for testing (default: 1000)')
    parser.add_argument('--epochs', type=int, default=14, metavar='N',
                        help='number of epochs to train (default: 14)')
    parser.add_argument('--lr', type=float, default=0.1, metavar='LR',
                        help='learning rate (default: 1.0)')
    parser.add_argument('--gamma', type=float, default=0.7, metavar='M',
                        help='Learning rate step gamma (default: 0.7)')
    parser.add_argument('--no-cuda', action='store_true', default=False,
                        help='disables CUDA training')
    parser.add_argument('--dry-run', action='store_true', default=False,
                        help='quickly check a single pass')
    parser.add_argument('--seed', type=int, default=1, metavar='S',
                        help='random seed (default: 1)')
    parser.add_argument('--log-interval', type=int, default=10, metavar='N',
                        help='how many batches to wait before logging training status')
    parser.add_argument('--save-model', action='store_true', default=True,
                        help='For Saving the current Model')
    args = parser.parse_args()
    use_cuda = not args.no_cuda and torch.cuda.is_available()

    torch.manual_seed(args.seed)

    device = torch.device("cuda" if use_cuda else "cpu")

    kwargs = {'batch_size': args.batch_size}
    if use_cuda:
        kwargs.update({'num_workers': 1,
                       'pin_memory': True,
                       'shuffle': True},
                     )

    transform=transforms.Compose([
        transforms.ToTensor(),
        transforms.Normalize((0.1307,), (0.3081,))
        ])
    dataset1 = datasets.MNIST('', train=True, download=False,
                       transform=transform)
    dataset2 = datasets.MNIST('', train=False,
                       transform=transform)
    train_loader = torch.utils.data.DataLoader(dataset1,**kwargs)
    test_loader = torch.utils.data.DataLoader(dataset2, **kwargs)

    model = Net().to(device)
    optimizer = optim.Adadelta(model.parameters(), lr=args.lr)
    #optimizer = optim.Adam(model.parameters())

    scheduler = StepLR(optimizer, step_size=1, gamma=args.gamma)
    for epoch in range(1, args.epochs + 1):
        train(args, model, device, train_loader, optimizer, epoch)
        test(model, device, test_loader)
        scheduler.step()


    if args.save_model:
        torch.save(model.state_dict(), "mnist_cnn.pt")

    model.load_state_dict(torch.load('mnist_cnn.pt'))

    #觀察測試結果
    for i in range(5):
        x, y = next(iter(test_loader))
        x,y = x.to(device),y.to(device)
        out = model(x)
        pred = out.argmax(dim=1)
        plot_image(x.cpu(), pred.cpu(), 'test')




if __name__ == '__main__':
    main()



#使用Adadelta 設置lr衰減
#Test set: Average loss: 1.4739, Accuracy: 9873/10000 (99%)

#使用SGD優化器，learning rate0.1 ，未設置lr的衰減
#Test set: Average loss: 1.4735, Accuracy: 9880/10000 (99%)

#使用Adam優化器，lr默認使用Adam的默認值0.001（使用0.1loss下不來） 未設置lr的衰減
#Test set: Average loss: 1.4749, Accuracy: 9862/10000 (99%)