Nested Learning Paradigm

Nested learning treats continual learning as a set of smaller nested optimization problems that self-modify to mitigate forgetting in local models.