Compacting, Picking and Growing for Unforgetting Continual Learning

Continual lifelong learning is essential to many applications. In this paper, we propose a simple but effective approach to continual deep learning.