Papers of the day   All papers

Functional Regularisation for Continual Learning


Jun 18 2019 DeepMind

Neural networks suffer from catastrophic forgetting when tasks are encountered sequentially. We overcome this by Bayesian inference in function space, using inducing point sparse GP methods and by optimising over rehearsal data points:
8 replies, 882 likes

Nov 27 2019 Jonathan Schwarz

We've updated our paper on Continual Learning with Gaussian Processes with some new ideas on task-boundary detection and additional ablation studies around inducing point optimisation:
0 replies, 128 likes

Dec 21 2019 Jonathan Schwarz

Happy to announce this has been accepted to ICLR 2020! See you all in Ethiopia.
2 replies, 48 likes