Gradient-based Editing of Memory Examples for Online Task-free Continual Learning

TL;DR

This paper introduces GMED, a gradient-based method for editing stored memory examples in continual learning, which enhances replay effectiveness and reduces catastrophic forgetting without explicit task boundaries.

Contribution

We propose GMED, a novel framework for editing memory examples via gradient updates to improve continual learning performance.

Findings

GMED improves performance on five out of six datasets.

Seamless integration with existing memory-based CL algorithms.

Significant outperformance of baselines and previous state-of-the-art.

Abstract

We explore task-free continual learning (CL), in which a model is trained to avoid catastrophic forgetting in the absence of explicit task boundaries or identities. Among many efforts on task-free CL, a notable family of approaches are memory-based that store and replay a subset of training examples. However, the utility of stored seen examples may diminish over time since CL models are continually updated. Here, we propose Gradient based Memory EDiting (GMED), a framework for editing stored examples in continuous input space via gradient updates, in order to create more "challenging" examples for replay. GMED-edited examples remain similar to their unedited forms, but can yield increased loss in the upcoming model updates, thereby making the future replays more effective in overcoming catastrophic forgetting. By construction, GMED can be seamlessly applied in conjunction with other memory-based CL algorithms to bring further improvement. Experiments validate the effectiveness of GMED, and our best method significantly outperforms baselines and previous state-of-the-art on five out of six datasets. Code can be found at https://github.com/INK-USC/GMED.