动态调整低秩适配器的秩，缓解持续学习中的灾难性遗忘。
Parameter Efficient Continual Learning with Dynamic Low-Rank Adaptation
- 根据任务与参考权重的参数距离，动态分配低秩适配器的秩。
- 在多个视觉模型和场景中显著优于现有基线方法。
- 无需记忆回放，轻量级训练，适合资源受限的持续学习应用。
灾难性遗忘是深度神经网络在持续学习（CL）中面临的核心挑战，会破坏已习得知识。参数高效微调的CL技术因其在轻量级训练下有效缓解遗忘而受到关注。然而，现有方法中的低秩适配器（LoRA）对秩的选择极为敏感，易导致资源分配不当和性能下降。为此，我们提出PEARL，一种无记忆回放的持续学习框架，通过动态分配LoRA组件的秩来应对这一问题。具体而言，PEARL利用参考任务权重，基于当前任务与参考权重在参数空间中的相似度，自适应地确定任务专属LoRA组件的秩。为验证PEARL的通用性，我们在三种视觉架构（ResNet、可分离卷积网络、视觉变换器）及多种持续学习场景中进行了评估，结果表明，PEARL在所有测试场景中均显著优于现有基线方法。
原文摘要 · Abstract (English)
Catastrophic forgetting has remained a critical challenge for deep neural networks in Continual Learning (CL) as it undermines consolidated knowledge when learning new tasks. Parameter efficient fine tuning CL techniques are gaining traction for their effectiveness in addressing catastrophic forgetting with a lightweight training schedule while avoiding degradation of consolidated knowledge in pre-trained models. However, low rank adapters (LoRA) in these approaches are highly sensitive to rank selection which can lead to sub-optimal resource allocation and performance. To this end, we introduce PEARL, a rehearsal-free CL framework that entails dynamic rank allocation for LoRA components during CL training. Specifically, PEARL leverages reference task weights and adaptively determines the rank of task-specific LoRA components based on the current tasks' proximity to reference task weights in parameter space. To demonstrate the versatility of PEARL, we evaluate it across three vision architectures (ResNet, Separable Convolutional Network and Vision Transformer) and a multitude of CL scenarios, and show that PEARL outperforms all considered baselines by a large margin.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。