Theconventional closed-set DAmethods generally assume that the source and target domains share the same label space. However, this assumption is often not realistic in practice.

Typically, DNNs suffer from drastic performance degradation of previously learned tasksafterlearning newknowledge, which isawell-documented phenomenon, knownascatastrophic forgetting [8,9,10].

It is regarded as a promising distributed and privacy-preserving method.

FL further makes efforts to adapt real-world scenarios, i.e., adapting

Figure 1: Our approach--LoCo--offers memory-efficient learning of location-consistent (LoCo) features.

We establish a new theoretical framework for learning under multi-class, instance-dependent label noise. This framework casts learning with label noise as a form of domain adaptation, in particular, domain adaptation under posterior drift.