在这个教程，你将学习如何通过迁移学习训练神经网络。你可以在 cs231n notes 了解更多关于迁移学习的内容。
引用这些笔记 实践中，很少有人从头开始训练整个卷积网络，因为拥有足够大小的数据集是比较少见的。替代的是, 通常会从一个大的数据集(例如 ImageNet, 包含120万的图片和1000个分类)预训练一个卷积网络, 然后将这个卷积网络作为初始化的网络, 或者是感兴趣任务的固定的特征提取器。
如下是两种主要的迁移学习的使用场景:
- 微调卷积网络: 取代随机初始化网络, 我们从一个预训练的网络初始化, 比如从 imagenet 1000 数据集预训练的网络. 其余的训练就像往常一样.
- 卷积网络作为固定的特征提取器: 在这里, 我们固定网络中的所有权重, 最后的全连接层除外. 最后的全连接层被新的随机权重替换, 并且, 只有这一层是被训练的.
# License: BSD
# Author: Sasank Chilamkurthy
from __future__ import print_function, division
import torch
import torch.nn as nn
import torch.optim as optim
from torch.optim import lr_scheduler
import numpy as np
import torchvision
from torchvision import datasets, models, transforms
import matplotlib.pyplot as plt
import time
import os
import copy
plt.ion() # interactive mode一、加载数据
我们将使用torchvision和torch.utils.data包来加载数据。
今天我们要解决的问题是训练一个模型来区分 ants (蚂蚁) 和 bees (蜜蜂)。
用于训练的 ants 和 bees 图片各120张。每一类用于验证的图片各75张。通常, 如果从头开始训练, 这个非常小的数据集不足以进行泛化。但是, 因为我们使用迁移学习, 应该可以取得很好的泛化效果。
这个数据集是一个非常小的 imagenet 子集。
### 下载图片数据
import os
import os.path
import errno
url ='https://download.pytorch.org/tutorial/hymenoptera_data.zip'
filename='hymenoptera_data.zip'
def download(root):
'''
下载数据用于训练和测试的ants和bees的图片压缩包。
使用zipfile包减压压缩包。
'''
root = os.path.expanduser(root)
import zipfile
#下载图片压缩包到指定路径
download_url(url,root,filename)
#获得当前路径
cwd = os.getcwd()
path = os.path.join(root, filename)
tar = zipfile.ZipFile(path, "r")
#解压文件
tar.extractall(root)
tar.close()
#切换到当前工作路径
os.chdir(cwd)
def download_url(url, root, filename):
from six.moves import urllib
root = os.path.expanduser(root)
fpath = os.path.join(root, filename)
try:
os.makedirs(root)
except OSError as e:
if e.errno == errno.EEXIST:
pass
else:
raise
# downloads file
if os.path.isfile(fpath) :
print('使用已下载文件: ' + fpath)
else:
try:
print('下载 ' + url + ' 到 ' + fpath)
urllib.request.urlretrieve(url, fpath)
except:
if url[:5] == 'https':
url = url.replace('https:', 'http:')
print('Failed download. Trying https -> http instead.'
' Downloading ' + url + ' to ' + fpath)
urllib.request.urlretrieve(url, fpath)
download('./root')# Data augmentation and normalization for training
# Just normalization for validation
data_transforms = {
'train': transforms.Compose([
transforms.RandomResizedCrop(224),
transforms.RandomHorizontalFlip(),
transforms.ToTensor(),
transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])
]),
'val': transforms.Compose([
transforms.Resize(256),
transforms.CenterCrop(224),
transforms.ToTensor(),
transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])
]),
}
data_dir = './root/hymenoptera_data'
image_datasets = {x: datasets.ImageFolder(os.path.join(data_dir, x),
data_transforms[x])
for x in ['train', 'val']}
dataloaders = {x: torch.utils.data.DataLoader(image_datasets[x], batch_size=4,
shuffle=True, num_workers=4)
for x in ['train', 'val']}
dataset_sizes = {x: len(image_datasets[x]) for x in ['train', 'val']}
class_names = image_datasets['train'].classes
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")二、训练模型
现在, 让我们写一个通用的函数来训练模型. 这里, 我们将会举例说明:
- 调度学习率
- 保存最佳的学习模型
下面函数中, scheduler 参数是torch.optim.lr_scheduler 中的 LR scheduler 对象。
def train_model(model, criterion, optimizer, scheduler, num_epochs=25):
since = time.time()
best_model_wts = copy.deepcopy(model.state_dict())
best_acc = 0.0
for epoch in range(num_epochs):
print('Epoch {}/{}'.format(epoch, num_epochs - 1))
print('-' * 10)
# Each epoch has a training and validation phase
for phase in ['train', 'val']:
if phase == 'train':
scheduler.step()
model.train() # Set model to training mode
else:
model.eval() # Set model to evaluate mode
running_loss = 0.0
running_corrects = 0
# Iterate over data.
for inputs, labels in dataloaders[phase]:
inputs = inputs.to(device)
labels = labels.to(device)
# zero the parameter gradients
optimizer.zero_grad()
# forward
# track history if only in train
with torch.set_grad_enabled(phase == 'train'):
outputs = model(inputs)
_, preds = torch.max(outputs, 1)
loss = criterion(outputs, labels)
# backward + optimize only if in training phase
if phase == 'train':
loss.backward()
optimizer.step()
# statistics
running_loss += loss.item() * inputs.size(0)
running_corrects += torch.sum(preds == labels.data)
epoch_loss = running_loss / dataset_sizes[phase]
epoch_acc = running_corrects.double() / dataset_sizes[phase]
print('{} Loss: {:.4f} Acc: {:.4f}'.format(
phase, epoch_loss, epoch_acc))
# deep copy the model
if phase == 'val' and epoch_acc > best_acc:
best_acc = epoch_acc
best_model_wts = copy.deepcopy(model.state_dict())
print()
time_elapsed = time.time() - since
print('Training complete in {:.0f}m {:.0f}s'.format(
time_elapsed // 60, time_elapsed % 60))
print('Best val Acc: {:4f}'.format(best_acc))
# load best model weights
model.load_state_dict(best_model_wts)
return model1.显示部分图像
让我们显示一些训练中的图片, 以便了解数据增强。
def imshow(inp, title=None):
"""Imshow for Tensor."""
inp = inp.numpy().transpose((1, 2, 0))
mean = np.array([0.485, 0.456, 0.406])
std = np.array([0.229, 0.224, 0.225])
inp = std * inp + mean
inp = np.clip(inp, 0, 1)
plt.imshow(inp)
if title is not None:
plt.title(title)
plt.pause(0.001) # pause a bit so that plots are updated
# Get a batch of training data
inputs, classes = next(iter(dataloaders['train']))
# Make a grid from batch
out = torchvision.utils.make_grid(inputs)
imshow(out, title=[class_names[x] for x in classes])2.显示模型的预测结果
写一个处理少量图片, 并显示预测结果的通用函数。
def visualize_model(model, num_images=6):
was_training = model.training
model.eval()
images_so_far = 0
fig = plt.figure()
with torch.no_grad():
for i, (inputs, labels) in enumerate(dataloaders['val']):
inputs = inputs.to(device)
labels = labels.to(device)
outputs = model(inputs)
_, preds = torch.max(outputs, 1)
for j in range(inputs.size()[0]):
images_so_far += 1
ax = plt.subplot(num_images//2, 2, images_so_far)
ax.axis('off')
ax.set_title('predicted: {}'.format(class_names[preds[j]]))
imshow(inputs.cpu().data[j])
if images_so_far == num_images:
model.train(mode=was_training)
return
model.train(mode=was_training)三、调整卷积网络
加载一个预训练的网络, 并重置最后一个全连接层。
model_ft = models.resnet18(pretrained=True)
num_ftrs = model_ft.fc.in_features
model_ft.fc = nn.Linear(num_ftrs, 2)
model_ft = model_ft.to(device)
criterion = nn.CrossEntropyLoss()
# Observe that all parameters are being optimized
optimizer_ft = optim.SGD(model_ft.parameters(), lr=0.001, momentum=0.9)
# Decay LR by a factor of 0.1 every 7 epochs
exp_lr_scheduler = lr_scheduler.StepLR(optimizer_ft, step_size=7, gamma=0.1)
训练和评估
CPU模式下将花费20—30分钟。在GPU环境下，花费时间少于1分钟（官方给的数据，我没有环境测试）。
model_ft = train_model(model_ft, criterion, optimizer_ft, exp_lr_scheduler,
num_epochs=25
visualize_model(model_ft)
四、卷积神经网络作为固定特征提取器
ConvNet as fixed feature extractor
这里, 我们固定网络中除最后一层外的所有权重. 为了固定这些参数, 我们需要设置 requires_grad == False , 然后在 backward() 中就不会计算梯度.
你可以在这里(http://t.cn/EafTQ8T)阅读更多相关信息.
model_conv = torchvision.models.resnet18(pretrained=True)
for param in model_conv.parameters():
param.requires_grad = False
# Parameters of newly constructed modules have requires_grad=True by default
num_ftrs = model_conv.fc.in_features
model_conv.fc = nn.Linear(num_ftrs, 2)
model_conv = model_conv.to(device)
criterion = nn.CrossEntropyLoss()
# Observe that only parameters of final layer are being optimized as
# opoosed to before.
optimizer_conv = optim.SGD(model_conv.fc.parameters(), lr=0.001, momentum=0.9)
# Decay LR by a factor of 0.1 every 7 epochs
exp_lr_scheduler = lr_scheduler.StepLR(optimizer_conv, step_size=7, gamma=0.1)训练和评估
在使用 CPU 的情况下, 和前一个方案相比, 这将花费的时间是它的一半。期望中, 网络的大部分是不需要计算梯度的. 前向传递依然要计算梯度。
model_conv = train_model(model_conv, criterion, optimizer_conv,
exp_lr_scheduler, num_epochs=25)
visualize_model(model_conv)
plt.ioff()
plt.show()
Type Markdown and LaTeX: α²
