Definition #
Applying knowledge from one domain to improve learning in another.
Key Characteristics #
- Fine-tuning
- Feature extraction
- Domain adaptation
Why It Matters #
Reduces training time and data requirements by 70-90%.
Examples #
- ImageNet-pretrained models
- Hugging Face Transformers