Online Continual Learning via Spiking Neural Networks with Sleep Enhanced Latent Replay

TL;DR

This paper introduces SESLR, a novel online continual learning method using spiking neural networks and sleep-inspired replay, significantly reducing memory use and bias while improving accuracy on multiple datasets.

Contribution

The paper proposes SESLR, a sleep-enhanced latent replay approach with SNNs that reduces memory overhead and mitigates bias in online continual learning.

Findings

Achieves nearly 30% accuracy improvement on CIFAR10 with one-third memory.

Improves accuracy by ~10% on CIFAR10-DVS while reducing memory by 32 times.

Demonstrates effectiveness on both conventional and neuromorphic datasets.

Abstract

Edge computing scenarios necessitate the development of hardware-efficient online continual learning algorithms to be adaptive to dynamic environment. However, existing algorithms always suffer from high memory overhead and bias towards recently trained tasks. To tackle these issues, this paper proposes a novel online continual learning approach termed as SESLR, which incorporates a sleep enhanced latent replay scheme with spiking neural networks (SNNs). SESLR leverages SNNs' binary spike characteristics to store replay features in single bits, significantly reducing memory overhead. Furthermore, inspired by biological sleep-wake cycles, SESLR introduces a noise-enhanced sleep phase where the model exclusively trains on replay samples with controlled noise injection, effectively mitigating classification bias towards new classes. Extensive experiments on both conventional (MNIST, CIFAR10) and neuromorphic (NMNIST, CIFAR10-DVS) datasets demonstrate SESLR's effectiveness. On Split CIFAR10, SESLR achieves nearly 30% improvement in average accuracy with only one-third of the memory consumption compared to baseline methods. On Split CIFAR10-DVS, it improves accuracy by approximately 10% while reducing memory overhead by a factor of 32. These results validate SESLR as a promising solution for online continual learning in resource-constrained edge computing scenarios.