Online-LoRA: Task-free Online Continual Learning via Low Rank Adaptation
This addresses the problem of memory constraints and privacy issues in rehearsal buffers for researchers and practitioners in continual learning, though it appears incremental as it builds on existing low-rank adaptation techniques.
The paper tackles catastrophic forgetting in online continual learning with non-stationary data streams by introducing Online-LoRA, a framework that finetunes pre-trained Vision Transformers using online weight regularization and loss dynamics for automatic data shift recognition, achieving better performance than state-of-the-art methods across multiple benchmark datasets.
Catastrophic forgetting is a significant challenge in online continual learning (OCL), especially for non-stationary data streams that do not have well-defined task boundaries. This challenge is exacerbated by the memory constraints and privacy concerns inherent in rehearsal buffers. To tackle catastrophic forgetting, in this paper, we introduce Online-LoRA, a novel framework for task-free OCL. Online-LoRA allows to finetune pre-trained Vision Transformer (ViT) models in real-time to address the limitations of rehearsal buffers and leverage pre-trained models' performance benefits. As the main contribution, our approach features a novel online weight regularization strategy to identify and consolidate important model parameters. Moreover, Online-LoRA leverages the training dynamics of loss values to enable the automatic recognition of the data distribution shifts. Extensive experiments across many task-free OCL scenarios and benchmark datasets (including CIFAR-100, ImageNet-R, ImageNet-S, CUB-200 and CORe50) demonstrate that Online-LoRA can be robustly adapted to various ViT architectures, while achieving better performance compared to SOTA methods. Our code will be publicly available at: https://github.com/Christina200/Online-LoRA-official.git.