用于无样本持续学习中漂移抵抗空间的LoRA减法
LoRA Subtraction for Drift-Resistant Space in Exemplar-Free Continual Learning
AI总结：
针对无样本持续学习中的特征漂移问题，提出漂移抵抗空间（DRS）和低秩适应减法（LoRA-）方法，通过从预训练权重中减去旧任务LoRA权重并结合三元组损失，有效平衡稳定性与可塑性，在多个数据集上取得最优结果。
摘要：
在持续学习（CL）中，灾难性遗忘通常由特征漂移引起。这一挑战在无样本持续学习（EFCL）设置中尤为突出，因为无法保留先前任务的样本，使得保存先验知识变得困难。为了解决这一问题，一些EFCL方法旨在识别在容纳新任务的同时最小化对先前任务影响的特征空间。然而，它们依赖于静态特征或从旧任务中存储的过时统计信息，这阻碍了它们捕获CL中特征空间的动态演变，导致性能随时间退化。在本文中，我们引入了漂移抵抗空间（DRS），它有效地处理特征漂移，而无需显式特征建模或存储先前任务。提出了一种称为低秩适应减法（LoRA-）的新型参数高效微调方法来开发DRS。该方法在处理新任务数据之前，从初始预训练权重中减去旧任务的LoRA权重，为模型训练建立DRS。因此，LoRA-增强了稳定性，提高了效率，并简化了实现。此外，通过使用三元组损失进行学习，稳定特征漂移允许更好的可塑性。我们的方法在多个数据集上始终取得最先进的结果，特别是对于长任务序列。
英文摘要：
In continual learning (CL), catastrophic forgetting often arises due to feature drift. This challenge is particularly prominent in the exemplar-free continual learning (EFCL) setting, where samples from previous tasks cannot be retained, making it difficult to preserve prior knowledge. To address this issue, some EFCL methods aim to identify feature spaces that minimize the impact on previous tasks while accommodating new ones. However, they rely on static features or outdated statistics stored from old tasks, which prevents them from capturing the dynamic evolution of the feature space in CL, leading to performance degradation over time. In this paper, we introduce the Drift-Resistant Space (DRS), which effectively handles feature drifts without requiring explicit feature modeling or the storage of previous tasks. A novel parameter-efficient fine-tuning approach called Low-Rank Adaptation Subtraction (LoRA-) is proposed to develop the DRS. This method subtracts the LoRA weights of old tasks from the initial pre-trained weight before processing new task data to establish the DRS for model training. Therefore, LoRA- enhances stability, improves efficiency, and simplifies implementation. Furthermore, stabilizing feature drifts allows for better plasticity by learning with a triplet loss. Our method consistently achieves state-of-the-art results, especially for long task sequences, across multiple datasets.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
