迁移学习
迁移学习(Transfer learning) 就是把已学训练好的模型参数迁移到新的模型来帮助新模型训练。考虑到大部分数据或任务是存在相关性的，所以通过迁移学习我们可以将已经学到的模型参数（也可理解为模型学到的知识）通过某种方式来分享给新模型从而加快并优化模型的学习效率不用像大多数网络那样从零学习。
本文使用VGG16模型用于迁移学习，最终得到一个能对猫狗图片进行辨识的CNN（卷积神经网络），测试集用来验证模型是否能够很好的工作。
猫狗分类器
本文使用迁移学习实现猫狗分类器。
数据集来自Kaggle比赛：Dogs vs. Cats Redux: Kernels Edition
利用pytorch实现迁移学习
首先进行图片的导入和预览
path = "dog_vs_cat"
transform = transforms.Compose([transforms.CenterCrop(224),
transforms.ToTensor(),
transforms.Normalize([0.5,0.5,0.5], [0.5,0.5,0.5])])
data_image = {x:datasets.ImageFolder(root = os.path.join(path,x),
transform = transform)
for x in ["train", "val"]}
data_loader_image = {x:torch.utils.data.DataLoader(dataset=data_image[x],
batch_size = 4,
shuffle = True)
for x in ["train", "val"]}输入的图片需要分辨率为224*224，所以使用transform.CenterCrop(224)对原始图片进行裁剪。载入的图片训练集合为20000个和验证集合为5000个，原始图片全部为训练集合，需自己拆分出一部分验证集合，输出的Label，1代表为狗，0代表猫。
X_train,y_train = next(iter(data_loader_image["train"]))
mean = [0.5, 0.5, 0.5]
std = [0.5, 0.5, 0.5]
img = torchvision.utils.make_grid(X_train)
img = img.numpy().transpose((1,2,0))
img = img*std + mean
print([classes[i] for i in y_train])
plt.imshow(img)
训练集的图片都为2242243。
迁移模型，打印出原始VGG模型结构为：
VGG(
(features): Sequential(
(0): Conv2d(3, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
(1): ReLU(inplace)
(2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
(3): ReLU(inplace)
(4): MaxPool2d(kernel_size=(2, 2), stride=(2, 2), dilation=(1, 1), ceil_mode=False)
(5): Conv2d(64, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
(6): ReLU(inplace)
(7): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
(8): ReLU(inplace)
(9): MaxPool2d(kernel_size=(2, 2), stride=(2, 2), dilation=(1, 1), ceil_mode=False)
(10): Conv2d(128, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
(11): ReLU(inplace)
(12): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
(13): ReLU(inplace)
(14): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
(15): ReLU(inplace)
(16): MaxPool2d(kernel_size=(2, 2), stride=(2, 2), dilation=(1, 1), ceil_mode=False)
(17): Conv2d(256, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
(18): ReLU(inplace)
(19): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
(20): ReLU(inplace)
(21): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
(22): ReLU(inplace)
(23): MaxPool2d(kernel_size=(2, 2), stride=(2, 2), dilation=(1, 1), ceil_mode=False)
(24): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
(25): ReLU(inplace)
(26): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
(27): ReLU(inplace)
(28): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
(29): ReLU(inplace)
(30): MaxPool2d(kernel_size=(2, 2), stride=(2, 2), dilation=(1, 1), ceil_mode=False)
)
(classifier): Sequential(
(0): Linear(in_features=25088, out_features=4096, bias=True)
(1): ReLU(inplace)
(2): Dropout(p=0.5)
(3): Linear(in_features=4096, out_features=4096, bias=True)
(4): ReLU(inplace)
(5): Dropout(p=0.5)
(6): Linear(in_features=4096, out_features=1000, bias=True)
)
)迁移过来的VGG16模型需适应新的需求，达到对猫狗图片很好的识别，因此改写VGG16的全连接层的最后一部分并且重新训练参数。
即使只是训练整个全连接层的全部参数，普通的电脑也会花费大量的时间，所以这里只训练全连接层的最后一层，就能达到很好的效果。
model.classifier = torch.nn.Sequential(torch.nn.Linear(25088, 4096),
torch.nn.ReLU(),
torch.nn.Dropout(p=0.5),
torch.nn.Linear(4096, 4096),
torch.nn.ReLU(),
torch.nn.Dropout(p=0.5),
torch.nn.Linear(4096, 2))
for parma in model.parameters():
parma.requires_grad = False
for index, parma in enumerate(model.classifier.parameters()):
if index == 6:
parma.requires_grad = True
if use_gpu:
model = model.cuda()
cost = torch.nn.CrossEntropyLoss()
optimizer = torch.optim.Adam(model.classifier.parameters())parma.requires_grid = False是冻结参数，即使发生新的训练也不会进行参数的更新。
这里还对全连接层的最后一层进行了改写，torch.nn.Linear(4096, 2)使得最后输出的结果只有两个，即只需要对猫狗进行分辨。
optimizer = torch.optim.Adam(model.classifier.parameters())只对全连接层参数进行更新优化，loss计算依然使用交叉熵。
对改写后的模型进行查看：
VGG(
(features): Sequential(
(0): Conv2d(3, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
(1): ReLU(inplace)
(2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
(3): ReLU(inplace)
(4): MaxPool2d(kernel_size=(2, 2), stride=(2, 2), dilation=(1, 1), ceil_mode=False)
(5): Conv2d(64, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
(6): ReLU(inplace)
(7): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
(8): ReLU(inplace)
(9): MaxPool2d(kernel_size=(2, 2), stride=(2, 2), dilation=(1, 1), ceil_mode=False)
(10): Conv2d(128, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
(11): ReLU(inplace)
(12): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
(13): ReLU(inplace)
(14): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
(15): ReLU(inplace)
(16): MaxPool2d(kernel_size=(2, 2), stride=(2, 2), dilation=(1, 1), ceil_mode=False)
(17): Conv2d(256, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
(18): ReLU(inplace)
(19): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
(20): ReLU(inplace)
(21): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
(22): ReLU(inplace)
(23): MaxPool2d(kernel_size=(2, 2), stride=(2, 2), dilation=(1, 1), ceil_mode=False)
(24): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
(25): ReLU(inplace)
(26): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
(27): ReLU(inplace)
(28): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
(29): ReLU(inplace)
(30): MaxPool2d(kernel_size=(2, 2), stride=(2, 2), dilation=(1, 1), ceil_mode=False)
)
(classifier): Sequential(
(0): Linear(in_features=25088, out_features=4096, bias=True)
(1): ReLU()
(2): Dropout(p=0.5)
(3): Linear(in_features=4096, out_features=4096, bias=True)
(4): ReLU()
(5): Dropout(p=0.5)
(6): Linear(in_features=4096, out_features=2, bias=True)
)
)这里我使用纯cpu进行训练，因为速度贼慢，我只进行了对100张图片进行训练的demo，进行1次训练的结果为：
Epoch0/1
----------
Batch 5, Train Loss:1.2027, Train ACC:90.0000
Batch 10, Train Loss:0.6853, Train ACC:92.0000
Batch 15, Train Loss:0.7109, Train ACC:91.0000
Batch 20, Train Loss:0.5332, Train ACC:93.0000
Batch 25, Train Loss:0.5215, Train ACC:94.0000
Batch 30, Train Loss:0.4346, Train ACC:95.0000
Batch 35, Train Loss:0.4213, Train ACC:95.0000
Batch 40, Train Loss:0.3748, Train ACC:95.0000
Batch 45, Train Loss:0.3541, Train ACC:95.0000
Batch 50, Train Loss:0.3501, Train ACC:94.0000
train Loss:0.3501, Correct:94.0000
val Loss:0.9151, Correct:88.0000
Training time is:6m 4s看到训练的Loss为0.3501， Accuraty准确率为94%。验证集的Loss为0.9151，Accuraty准确率为88%。这只是100张图片的一次训练，更加多的图片以及多次的训练可能会得到一个更加好的结果。
随机输入测试集合产看预测结果：
可以看到预测结果没有出现错误，本文输入时采用了随机裁剪，如果对原始图片进行缩放可能会提升模型的预测准确率，此外还可以增加数据个数、训练次数、数据增强处理。
本文章为转载内容，我们尊重原作者对文章享有的著作权。如有内容错误或侵权问题，欢迎原作者联系我们进行内容更正或删除文章。
