A way for models to learn from many tasks and adapt fast to new ones.
Meta-learning is like teaching a kid how to study, not just drilling spelling words. A new quiz lands, and they spot the trick fast.
It trains on many tasks, so it can handle a new task after a few examples. You see it in Few-Shot Learning and personal AI tools.
Few-Shot Learning
Meta-learning is often used to help Few-Shot Learning adapt faster.
Transfer Learning
Transfer Learning reuses knowledge, and meta-learning teaches how to reuse it well.
Multitask Learning
Multitask Learning gives meta-learning many tasks to practice on.