[R语言] 模型微调

• 在源数据集（如ImageNet数据集）上预训练一个神经网络模型，即源模型。
  
• 创建一个新的神经网络模型，即目标模型。它复制了源模型上除了输出层外的所有模型设计及其参数。我们假设这些模型参数包含了源数据集上学习到的知识，且这些知识同样适用于目标数据集。我们还假设源模型的输出层跟源数据集的标签紧密相关，因此在目标模型中不予采用。
  
• 为目标模型添加一个输出大小为目标数据集类别个数的输出层，并随机初始化该层的模型参数。
  
• 在目标数据集（如椅子数据集）上训练目标模型。我们将从头训练输出层，而其余层的参数都是基于源模型的参数微调得到的。
  

%matplotlib inline
import torch
from torch import nn, optim
from torch.utils.data import Dataset, DataLoader
import torchvision
from torchvision.datasets import ImageFolder
from torchvision import transforms
from torchvision import models
import os
import sys
sys.path.append("/home/kesci/input/")
import d2lzh1981 as d2l
os.environ["CUDA_VISIBLE_DEVICES"] = "0"
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')

import os
os.listdir('/home/kesci/input/resnet185352')

['resnet18-5c106cde.pth']

data_dir = '/home/kesci/input/hotdog4014'
os.listdir(os.path.join(data_dir, "hotdog"))

['test', 'train']

train_imgs = ImageFolder(os.path.join(data_dir, 'hotdog/train'))
test_imgs = ImageFolder(os.path.join(data_dir, 'hotdog/test'))

hotdogs = [train_imgs[i][0] for i in range(8)]
not_hotdogs = [train_imgs[-i - 1][0] for i in range(8)]
d2l.show_images(hotdogs + not_hotdogs, 2, 8, scale=1.4);

注: 在使用预训练模型时，一定要和预训练时作同样的预处理。
如果你使用的是torchvision的models，那就要求:
All pre-trained models expect input images normalized in the same way, i.e. mini-batches of 3-channel RGB images of shape (3 x H x W), where H and W are expected to be at least 224. The images have to be loaded in to a range of [0, 1] and then normalized using mean = [0.485, 0.456, 0.406] and std = [0.229, 0.224, 0.225].

normalize = transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
train_augs = transforms.Compose([
    transforms.RandomResizedCrop(size=224),
    transforms.RandomHorizontalFlip(),
    transforms.ToTensor(),
    normalize
  ])
test_augs = transforms.Compose([
    transforms.Resize(size=256),
    transforms.CenterCrop(size=224),
    transforms.ToTensor(),
    normalize
  ])

pretrained_net = models.resnet18(pretrained=False)
pretrained_net.load_state_dict(torch.load('/home/kesci/input/resnet185352/resnet18-5c106cde.pth'))

<All keys matched successfully>

print(pretrained_net.fc)

Linear(in_features=512, out_features=1000, bias=True)

注: 如果你使用的是其他模型，那可能没有成员变量fc（比如models中的VGG预训练模型），所以正确做法是查看对应模型源码中其定义部分，这样既不会出错也能加深我们对模型的理解。pretrained-models.pytorch仓库貌似统一了接口，但是我还是建议使用时查看一下对应模型的源码。

pretrained_net.fc = nn.Linear(512, 2)
print(pretrained_net.fc)

Linear(in_features=512, out_features=2, bias=True)

output_params = list(map(id, pretrained_net.fc.parameters()))
feature_params = filter(lambda p: id(p) not in output_params, pretrained_net.parameters())
lr = 0.01
optimizer = optim.SGD([{'params': feature_params},
             {'params': pretrained_net.fc.parameters(), 'lr': lr * 10}],
             lr=lr, weight_decay=0.001)

def train_fine_tuning(net, optimizer, batch_size=128, num_epochs=5):
  train_iter = DataLoader(ImageFolder(os.path.join(data_dir, 'hotdog/train'), transform=train_augs),
              batch_size, shuffle=True)
  test_iter = DataLoader(ImageFolder(os.path.join(data_dir, 'hotdog/test'), transform=test_augs),
               batch_size)
  loss = torch.nn.CrossEntropyLoss()
  d2l.train(train_iter, test_iter, net, loss, optimizer, device, num_epochs)

train_fine_tuning(pretrained_net, optimizer)

training on  cpu
epoch 1, loss 3.4516, train acc 0.687, test acc 0.884, time 298.2 sec
epoch 2, loss 0.1550, train acc 0.924, test acc 0.895, time 296.2 sec
epoch 3, loss 0.1028, train acc 0.903, test acc 0.950, time 295.0 sec
epoch 4, loss 0.0495, train acc 0.931, test acc 0.897, time 294.0 sec
epoch 5, loss 0.1454, train acc 0.878, test acc 0.939, time 291.0 sec

scratch_net = models.resnet18(pretrained=False, num_classes=2)
lr = 0.1
optimizer = optim.SGD(scratch_net.parameters(), lr=lr, weight_decay=0.001)
train_fine_tuning(scratch_net, optimizer)

training on  cpu
epoch 1, loss 2.6391, train acc 0.598, test acc 0.734, time 292.4 sec
epoch 2, loss 0.2703, train acc 0.790, test acc 0.632, time 289.7 sec
epoch 3, loss 0.1584, train acc 0.810, test acc 0.825, time 290.2 sec
epoch 4, loss 0.1177, train acc 0.805, test acc 0.787, time 288.6 sec
epoch 5, loss 0.0782, train acc 0.829, test acc 0.828, time 289.8 sec

training on  cuda
epoch 1, loss 2.6686, train acc 0.582, test acc 0.556, time 25.3 sec
epoch 2, loss 0.2434, train acc 0.797, test acc 0.776, time 25.3 sec
epoch 3, loss 0.1251, train acc 0.845, test acc 0.802, time 24.9 sec
epoch 4, loss 0.0958, train acc 0.833, test acc 0.810, time 25.0 sec
epoch 5, loss 0.0757, train acc 0.836, test acc 0.780, time 24.9 sec