Open Peer Review. Open Publishing. Open Access. Open Discussion. Open Directory. Open Recommendations. Open API. Open Source.
Overcoming Catastrophic Interference using Conceptor-Aided Backpropagation
Xu He, Herbert Jaeger
Feb 15, 2018 (modified: Feb 19, 2018)ICLR 2018 Conference Blind Submissionreaders: everyoneShow Bibtex
Abstract:Catastrophic interference has been a major roadblock in the research of continual learning. Here we propose a variant of the back-propagation algorithm, "Conceptor-Aided Backprop" (CAB), in which gradients are shielded by conceptors against degradation of previously learned tasks. Conceptors have their origin in reservoir computing, where they have been previously shown to overcome catastrophic forgetting. CAB extends these results to deep feedforward networks. On the disjoint and permuted MNIST tasks, CAB outperforms two other methods for coping with catastrophic interference that have recently been proposed.
TL;DR:We propose a variant of the backpropagation algorithm, in which gradients are shielded by conceptors against degradation of previously learned tasks.