1 paper
Dhrupad Bhardwaj, Julia Kempe, Artem Vysogorets +2
This work proposes a new method to sequentially train deep neural networks on multiple tasks without suffering catastrophic forgetting, while endowing it with the capability to qui…