Baidu Encyclopedia version
Migration learning is a machine learning method that refers to a pre-trained model being reused in another task. Migration learning is related to multitasking learning and concept drifting. It is not a specialized machine learning domain.
However, migration learning is very popular in certain deep learning problems, such as in the case of resources that have a large amount of training depth models or a large number of data sets used to pre-train the models. Migration learning works only when the depth model feature in the first task is a generalization feature.
This migration in deep learning is called inductive migration. It is by way of a model that is suitable for different but related tasks that narrows the search range of possible models in an advantageous way.
Migration learning is a research problem in machine learning that focuses on storing the knowledge gained in solving a problem and applying it to different but related problems. For example, the knowledge gained in learning to identify a car can be applied when trying to identify a truck. This area of research has a relationship with the long history of the psychology literature on transfer, although the formal relationship between the two domains is limited.