Unsupervised Replay Strategies for Continual Learning with Limited Data

TL;DR

This paper investigates how incorporating an unsupervised sleep phase with stochastic activation and Hebbian learning improves continual learning in neural networks trained on limited and imbalanced datasets, reducing forgetting and enhancing accuracy.

Contribution

It introduces a novel sleep replay strategy that significantly boosts continual learning performance and mitigates catastrophic forgetting in neural networks with scarce data.

Findings

Sleep replay improves accuracy with limited data.

Sleep rescues and enhances previously learned tasks.

Sleep reduces catastrophic forgetting in continual learning.

Abstract

Artificial neural networks (ANNs) show limited performance with scarce or imbalanced training data and face challenges with continuous learning, such as forgetting previously learned data after new tasks training. In contrast, the human brain can learn continuously and from just a few examples. This research explores the impact of 'sleep', an unsupervised phase incorporating stochastic activation with local Hebbian learning rules, on ANNs trained incrementally with limited and imbalanced datasets, specifically MNIST and Fashion MNIST. We discovered that introducing a sleep phase significantly enhanced accuracy in models trained with limited data. When a few tasks were trained sequentially, sleep replay not only rescued previously learned information that had been catastrophically forgetting following new task training but often enhanced performance in prior tasks, especially those trained with limited data. This study highlights the multifaceted role of sleep replay in augmenting learning efficiency and facilitating continual learning in ANNs.