Overcoming Catastrophic Forgetting by Generative Regularization

By Patrick H. Chen, Wei Wei, Cho-Jui Hsieh, Bo Dai. 2019

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

1. Matteo Boschini
2. Liyuan Wang
3. Gyuhak Kim
4. Yoshua Bengio
5. Angelo Porrello
6. Dacheng Tao
7. Zhen Wang
8. Sergey Levine

Citations in the corpus

• Embracing Change: Continual Learning in Deep Neural Networks. (2020, methodology): “These regularization-based methods estimate the importance of each model parameter for previous tasks and penalize changes to each parameter proportional to this measure (Figure 1B) [5,24 – 28].”
• Rethinking Experience Replay: a Bag of Tricks for Continual Learning (2020, methodology): “On the Split Fashion-MNIST setting, we apply all evaluated CL methods on a fully-connected network with two hidden layers of 256 ReLU units each, in line with [13], [14].”