Transfer Learning
Reusing a model trained on one big task as the starting point for a different, usually smaller task. The features it already learned give the new task a huge head start.
Think of It Like This
A guitarist picking up a mandolin — most of the musical skill carries straight over.
Instead of training from random weights, you take a model that already learned general features — edges and textures for images, grammar and meaning for text — and adapt it to your problem. That's why a strong image classifier can be retrained on just a few thousand of your own photos and still work well. Fine-tuning is the most common way to do the adapting, and convolutional networks were where the trick first went mainstream.