A technique where knowledge learned from one task or domain is applied to a different but related task, reducing the data and compute needed for the new task.
When AI uses what it learned from one thing to help with something new — like how learning to ride a bike helps you learn a motorcycle.
The idea that an AI trained on one task can apply that knowledge to a different task — so you don't have to start from scratch every time.
A technique where knowledge learned from one task or domain is applied to a different but related task, reducing the data and compute needed for the new task.
Leveraging representations learned during pre-training on a source task to improve performance on a target task — the fundamental paradigm behind foundation models and their downstream adaptations.
The exploitation of shared structure between source and target domains via learned representations — formalized as minimizing target risk under domain shift, with theoretical bounds governed by the divergence between source and target distributions.
Want to explore Transfer Learning in depth?
Ask SeekBox and get answers from 7 AI engines at once.
Try it in SeekBox →