Transfer Learning for Images Using PyTorch: Essential Training

Discover how to implement transfer learning using PyTorch, the popular machine learning framework.

Introduction

• Welcome
• What you should know before watching this course

1. What Is Transfer Learning?

• What is transfer learning?
• VGG16
• CIFAR-10 dataset

2. Transfer Learning: Fixed Feature Extractor

• Creating a fixed feature extractor
• Understanding loss: CrossEntropyLoss() and NLLLoss()
• Autograd
• Using autograd
• Training the fixed feature extractor
• Optimizers
• CPU to GPU
• Train the extractor
• Evaluate the network and viewing images
• Viewing images and normalization
• Accuracy of the model

3. Fine-Tuning the ConvNet

• Fine-tuning
• Using fine-tuning
• Training from the fully connected network onwards
• Unfreezing and training over the last CNN block onwards
• Unfreezing and training over the last two CNN block onwards

4. Further Techniques

• Learning rates
• Differential learning rates

Conclusion

• Next steps