Transfer Learning and domain adaptation refer to the situation where what has been learned in one setting… is expoilted to improve generalization in another setting. Is the improvement of learning a new task through thew transfer of knowledge from a related task that has already been learned.
Learning a new task relies on the previous learned task
Terminology:
- source domain the previous learned task
- target domain the new task
Transductive Transfer Learning
Refers to the situations where the label information only comes from the source domain, it the task is the same but the domain are different (still related e.g image/audio of animals) it’s called domain adaptation
Inductive Transfer Learning
We can have source and target label, in this case it’s called multi-task learning, or just the target labels self-taught learning.
Unsupervised transfer Learning
No labels on both source and target, clustering and dimensionality reduction.