no code implementations • 29 Sep 2021 • Lawrence Ki-On Chan, James Kwok
In this paper, we remove this inconsistency in the use of ER and improve continual learning representations by integrating ER also into meta-training.
Continual Learning Meta-Learning