Maintaining Plasticity in Continual Learning via Regenerative Regularization

This paper proposes a conceptually simple and easy to implement method for mitigating plasticity loss in continual learning.