Transfer Learning

Page 1

What id transfer Learning?

The knowledge from learning in one setting is reused to improve learning in another setting. Transfer learning is inspired by the way human learners take advantage of their existing knowledge and skills: a human who knows how to read literature is more likely to succeed in reading scientific papers than a human who does not know how to read at all. In the context of supervised learning, transfer learning implies the ability to reuse the knowledge of the dependence structure between features and labels learned in one setting to improve the inference of the dependence structure in another setting.

Page 2

Page 3

instead of learning in one setting, the knowledge from learning in one setting is reused to improve learning in another setting. Transfer learning is inspired by the way human learners take advantage of their existing knowledge and skills: a human who knows how to read literature is more likely to succeed in reading scientific papers than a human who does not know how to read at all. In the context of supervised learning, transfer learning implies the ability to reuse the knowledge of the dependence structure between features and labels learned in one setting to improve the inference of the dependence structure in another setting.