Learning to Continually Learn Rapidly from Few and Noisy Data

TL;DR

This paper introduces a meta-learning approach to continual learning that enables neural networks to rapidly adapt to new tasks with limited memory and noisy data, effectively reducing catastrophic forgetting.

Contribution

It proposes a meta-learning method that learns a per-parameter learning rate for each past task, improving continual learning with less memory and noisy data.

Findings

Achieved strong results with limited memory resources.

Demonstrated robustness to noisy data in continual learning.

Enabled faster adaptation with fewer updates.

Abstract

Neural networks suffer from catastrophic forgetting and are unable to sequentially learn new tasks without guaranteed stationarity in data distribution. Continual learning could be achieved via replay -- by concurrently training externally stored old data while learning a new task. However, replay becomes less effective when each past task is allocated with less memory. To overcome this difficulty, we supplemented replay mechanics with meta-learning for rapid knowledge acquisition. By employing a meta-learner, which \textit{learns a learning rate per parameter per past task}, we found that base learners produced strong results when less memory was available. Additionally, our approach inherited several meta-learning advantages for continual learning: it demonstrated strong robustness to continually learn under the presence of noises and yielded base learners to higher accuracy in less updates.