1.Cross Domain
Cross Domain: Translated into Chinese as cross-domain. Taking small-sample learning as an example, the cross-domain problem refers to the fact that during the learning process, there is a difference in one or more items in the feature space, category space, or marginal distribution between the source domain and the target domain. problems arising during the learning process .
2.Domain Adaptation
Abbreviation DA: Chinese representation--domain adaptation, mainly for the solution of the source domain and the target domain in the Cross Domain cross-domain problem, the feature space and category space are the same, but the edge distribution is different, as shown in the figure below; domain adaptation is more The application is UDA---Unsupervised Domain Adaptation under unsupervised learning . There has been a long history of research on domain adaptation techniques, which aim to transfer knowledge from one or more source domains to target domains with different data distributions .
3.Domain Generalization
Domain Generalization, abbreviated as DG, is translated into domain generalization in Chinese, which is mainly a solution for more complex source domains and target domains with different category spaces, as shown in the figure below.
