Overcoming catastrophic forgetting in neural networks 2017

Back to Research-Papers

Previously, forgetting was seen as an intrinsic part of neural networks (task to task). This paper shows that it is possible to overcome this limitation