What are the most popular domain adaptation methods (for transfer learning)?

Question

I understand supervised and unsupervised learning well, and would be able to identify some 'basic' examples of, for example, supervised classifcation as:

• SVMs
• Random Forests
• Logistic Regression

These are key works in the field which have lots of code and publications available.

I am now starting to look at domain adaptation in supervised learning, where the distribution over the data at learning and testing time are known to be different. Despite reading some of the literature, I haven't spotted any similar 'basic' methods which come up time and time again. In contrast, there seem to be a wild array of completely different methods for achieving the goal, many of which have only been in the literature for a few years.

Are there such key, established methods for domain adaptation? What are the most popular methods currently used?

Answer 1

Domain Adaptation is the process that attempts to alter the source domain in a way to bring the distribution of the source closer to that of the target domain. In many cases the Domain Adaptation methods are modifications of the basic algorithms from the area of the traditional Machine Learning, that they are trying to take into account the difference between the distributions of the train and the test set. For example, one popular DA algorithm is TrAdaboost, which is a Boosting algorithm and it is a modification of the Adaboost algorithm. Another issue is that the DA methods are developed based on the kind of the difference between the two domains that they are trying to tackle. Some of them are trying to get closer the marginal distributions and others the conditional distributions.

In my opinion, they do not exist basic or popular methods in same sense as in the traditional Machine Learning, because the DA methods are strongly related with the application or with the assumption that they are based on. You can get an idea of the state of the art of the existing apporaches studying these surveys:

PAN, S. J.; YANG, Q. A Survey on Transfer Learning. IEEE Transactions on Knowledge and Data Engineering, v. 22, n. 10, p. 1345–1359, out. 2010.

Weiss, K., Khoshgoftaar, T.M. & Wang: A Survey of Transfer Learning. D. J Big Data, 2016.