Transfer Learning for Deep Learning

Transfer Learning for Deep Learning with PyTorch (Alien vs. Predator)

What is Transfer Learning?

Transfer learning is a technique of using a trained model to solve another related task. It's popular to use other network model weight to reduce your training time because you need a lot of data to train a network model. To reduce the training time, you use other network and its weight and modify the last layer to solve our problem. The advantage is you can use a small dataset to train the last layer.

Loading Dataset

Source: Alien vs. Predator Kaggle

Before you start, you need to understand the dataset that you are going to use. In this part, you will classify an Alien and a Predator from nearly 700 images. For this technique, you don't really need a big amount of data to train. You can download the dataset from Kaggle: Alien vs. Predator.

Step 1) Load the Data

The first step is to load our data and do some transformation to images so that it matched the network requirements. You will load the data from a folder with torchvision.dataset. The module will iterate in the folder to split the data for train and validation. The transformation process will crop the images from the center, perform a horizontal flip, normalize, and finally convert it to tensor.

from __future__ import print_function, division
import os
import time
import torch
import torchvision
from torchvision import datasets, models, transforms
import torch.optim as optim
import numpy as np
import matplotlib.pyplot as plt

data_dir = "alien_pred"
input_shape = 224
mean = [0.5, 0.5, 0.5]
std = [0.5, 0.5, 0.5]

#data transformation
data_transforms = {
   'train': transforms.Compose([
       transforms.CenterCrop(input_shape),
       transforms.ToTensor(),
       transforms.Normalize(mean, std)
   ]),
   'validation': transforms.Compose([
       transforms.CenterCrop(input_shape),
       transforms.ToTensor(),
       transforms.Normalize(mean, std)
   ]),
}

image_datasets = {
   x: datasets.ImageFolder(
       os.path.join(data_dir, x),
       transform=data_transforms[x]
   )
   for x in ['train', 'validation']
}

dataloaders = {
   x: torch.utils.data.DataLoader(
       image_datasets[x], batch_size=32,
       shuffle=True, num_workers=4
   )
   for x in ['train', 'validation']
}

dataset_sizes = {x: len(image_datasets[x]) for x in ['train', 'validation']}

print(dataset_sizes)
class_names = image_datasets['train'].classes

device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")

Let's visualize our dataset. The visualization process will get the next batch of images from the train data-loaders and labels and display it with matplot.

images, labels = next(iter(dataloaders['train']))

rows = 4
columns = 4
fig=plt.figure()
for i in range(16):
   fig.add_subplot(rows, columns, i+1)
   plt.title(class_names[labels[i]])
   img = images[i].numpy().transpose((1, 2, 0))
   img = std * img + mean
   plt.imshow(img)
plt.show()

Step 2) Define Model

In this process, you will use ResNet18 from torchvision module. You will use torchvision.models to load resnet18 with the pre-trained weight set to be True. After that, you will freeze the layers so that these layers are not trainable. You also modify the last layer with a Linear layer to fit with our needs that is 2 classes. You also use CrossEntropyLoss for multi-class loss function and for the optimizer you will use SGD with the learning rate of 0.0001 and a momentum of 0.9.

## Load the model based on VGG19
vgg_based = torchvision.models.vgg19(pretrained=True)

## freeze the layers
for param in vgg_based.parameters():
   param.requires_grad = False

# Modify the last layer
number_features = vgg_based.classifier[6].in_features
features = list(vgg_based.classifier.children())[:-1] # Remove last layer
features.extend([torch.nn.Linear(number_features, len(class_names))])
vgg_based.classifier = torch.nn.Sequential(*features)

vgg_based = vgg_based.to(device)

print(vgg_based)

criterion = torch.nn.CrossEntropyLoss()
optimizer_ft = optim.SGD(vgg_based.parameters(), lr=0.001, momentum=0.9)

The output model structure

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, stride=2, padding=0, dilation=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, stride=2, padding=0, dilation=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): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
	(17): ReLU(inplace)
	(18): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
	(19): Conv2d(256, 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): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
	(24): ReLU(inplace)
	(25): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
	(26): ReLU(inplace)
	(27): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
	(28): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
	(29): ReLU(inplace)
	(30): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
	(31): ReLU(inplace)
	(32): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
	(33): ReLU(inplace)
	(34): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
	(35): ReLU(inplace)
	(36): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=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=2, bias=True)
  )
)

Step 3) Train and Test Model

We will use some of the function from PyTorch Tutorial to help us train and evaluate our model.

def train_model(model, criterion, optimizer, num_epochs=25):
   since = time.time()

   for epoch in range(num_epochs):
       print('Epoch {}/{}'.format(epoch, num_epochs - 1))
       print('-' * 10)

       #set model to trainable
       # model.train()

       train_loss = 0

       # Iterate over data.
       for i, data in enumerate(dataloaders['train']):
           inputs , labels = data
           inputs = inputs.to(device)
           labels = labels.to(device)

           optimizer.zero_grad()
          
           with torch.set_grad_enabled(True):
               outputs  = model(inputs)
               loss = criterion(outputs, labels)

           loss.backward()
           optimizer.step()

           train_loss += loss.item() * inputs.size(0)

           print('{} Loss: {:.4f}'.format(
               'train', train_loss / dataset_sizes['train']))
          
   time_elapsed = time.time() - since
   print('Training complete in {:.0f}m {:.0f}s'.format(
       time_elapsed // 60, time_elapsed % 60))

   return model

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['validation']):
           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: {} truth: {}'.format(class_names[preds[j]], class_names[labels[j]]))
               img = inputs.cpu().data[j].numpy().transpose((1, 2, 0))
               img = std * img + mean
               ax.imshow(img)

               if images_so_far == num_images:
                   model.train(mode=was_training)
                   return
       model.train(mode=was_training)

Finally, let's start our training process with the number of epochs set to 25 and evaluate after the training process. At each training step, the model will take the input and predict the output. After that, the predicted output will be passed to the criterion to calculate the losses. Then the losses will perform a backprop calculation to calculate the gradient and finally calculating the weights and optimize the parameters with autograd.

At the visualize model, the trained network will be tested with a batch of images to predict the labels. Then it will be visualized with the help of matplotlib.

vgg_based = train_model(vgg_based, criterion, optimizer_ft, num_epochs=25)

visualize_model(vgg_based)

plt.show()

Results

The final result is that you achieved an accuracy of 92%.

Epoch 23/24
----------
train Loss: 0.0044
train Loss: 0.0078
train Loss: 0.0141
train Loss: 0.0221
train Loss: 0.0306
train Loss: 0.0336
train Loss: 0.0442
train Loss: 0.0482
train Loss: 0.0557
train Loss: 0.0643
train Loss: 0.0763
train Loss: 0.0779
train Loss: 0.0843
train Loss: 0.0910
train Loss: 0.0990
train Loss: 0.1063
train Loss: 0.1133
train Loss: 0.1220
train Loss: 0.1344
train Loss: 0.1382
train Loss: 0.1429
train Loss: 0.1500
Epoch 24/24
----------
train Loss: 0.0076
train Loss: 0.0115
train Loss: 0.0185
train Loss: 0.0277
train Loss: 0.0345
train Loss: 0.0420
train Loss: 0.0450
train Loss: 0.0490
train Loss: 0.0644
train Loss: 0.0755
train Loss: 0.0813
train Loss: 0.0868
train Loss: 0.0916
train Loss: 0.0980
train Loss: 0.1008
train Loss: 0.1101
train Loss: 0.1176
train Loss: 0.1282
train Loss: 0.1323
train Loss: 0.1397
train Loss: 0.1436
train Loss: 0.1467
Training complete in 2m 47s

End then the output of our model will be visualized with matplot below:

Summary

So, let's summarize everything! The first factor is PyTorch is a growing deep learning framework for beginners or for research purpose. It offers high computation time, Dynamic Graph, GPUs support and it's totally written in Python. You are able to define our own network module with ease and do the training process with an easy iteration. It's clear that PyTorch is ideal for beginners to find out deep learning and for professional researchers it's very useful with faster computation time and also the very helpful autograd function to assist dynamic graph.

Reference : https://www.guru99.com/transfer-learning.html