GRASP: A Rehearsal Policy for Efficient Online Continual Learning

arXiv:2308.13646v2 Announce Type: replace
Abstract: Continual learning (CL) in deep neural networks (DNNs) involves incrementally accumulating knowledge in a DNN from a growing data stream. A major challenge in CL is that non-stationary data streams cause catastrophic forgetting of previously learned abilities. A popular solution is rehearsal: storing past observations in a buffer and then sampling the buffer to update the DNN. Uniform sampling in a class-balanced manner is highly effective, and better sample selection policies have been elusive. Here, …

abstract arxiv catastrophic forgetting challenge continual cs.cl cs.cv cs.lg data data stream data streams dnn knowledge major networks neural networks policy popular solution type