Introduction
If you already play the guitar, learning the ukulele is much easier: you don't start from zero — your knowledge of chords, rhythm and finger movement transfers.
Transfer learning does the same for models: take a model that was already trained on a big, general task (the source), and reuse its knowledge for your task (the target), which often has far less data.
It's the reason modern AI is practical:
- an image model pre-trained on millions of photos can learn to recognise your 5 product types from a few hundred images;
- a large language model pre-trained on huge amounts of text can be adapted to classify your support tickets with a few thousand examples;
- training from scratch would need far more data, time and money.