Transfer Learning

What Transfer Learning is

Rather than learning from scratch, you inherit general representations and adapt them, which is why a few hundred labelled examples can now be enough for a workable classifier.

How it works

A pretrained backbone is either frozen and used as a feature extractor, or fine-tuned end-to-end at a low learning rate on the target data.

Why it matters

It is the economic engine of modern AI: pretraining is done once at enormous cost, and thousands of downstream applications adapt it cheaply.

Common uses

• →Domain-specific classifiers
• →Fine-tuned language models
• →Medical imaging with small datasets