arXiv 2408.149762024-08-27cs.LGcs.AIcs.CV

无先验平衡回放：面向长尾持续学习的不确定性引导蓄水池采样

Prior-free Balanced Replay: Uncertainty-guided Reservoir Sampling for Long-Tailed Continual Learning

Shenzhen Research Institute of Big Data (SRIBD)（深圳大数据研究院）；The Chinese University of Hong Kong, Shenzhen（香港中文大学（深圳））；The Hong Kong University of Science and Technology (Guangzhou)（香港科技大学（广州））；The Hong Kong University of Science and Technology（香港科技大学）；The Hong Kong Polytechnic University（香港理工大学）

Lei Liu, Li Liu, Yawen Cui

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

针对长尾持续学习中的灾难性遗忘问题，提出无先验平衡回放（PBR）框架，通过不确定性引导的蓄水池采样及边界与原型约束，在无需先验信息下有效减少遗忘并超越现有方法。

摘要：

即使在大型模型时代，持续学习（CL）中一个众所周知的问题依然是灾难性遗忘，当持续数据流呈现长尾分布时，这一问题极具挑战性，被称为长尾持续学习（LTCL）。现有的LTCL解决方案通常需要数据流的标签分布来实现重平衡训练。然而，在真实场景中获取此类先验信息往往是不可行的，因为模型应该在不预先识别多数类和少数类的情况下进行学习。为此，我们提出了一种新颖的无先验平衡回放（PBR）框架，以从长尾数据流中学习并减少遗忘。具体而言，受我们的实验发现启发——即少数类由于具有更高的不确定性而更容易被遗忘，我们新设计了一种不确定性引导的蓄水池采样策略，该策略基于模型与样本之间的相互依赖关系，在不使用任何先验信息的情况下优先重演少数类数据。此外，我们引入了两个无先验组件以进一步缓解遗忘问题：（1）边界约束，用于保留不确定的边界支持样本，以持续重新估计任务边界；（2）原型约束，用于在训练过程中保持已学习类原型的一致性。我们的方法在三个标准长尾基准上进行了评估，结果表明，在任务增量和类增量学习设置，以及有序和打乱的LTCL设置中，其性能均优于现有的CL方法和此前的SOTA LTCL方法。

英文摘要：

Even in the era of large models, one of the well-known issues in continual learning (CL) is catastrophic forgetting, which is significantly challenging when the continual data stream exhibits a long-tailed distribution, termed as Long-Tailed Continual Learning (LTCL). Existing LTCL solutions generally require the label distribution of the data stream to achieve re-balance training. However, obtaining such prior information is often infeasible in real scenarios since the model should learn without pre-identifying the majority and minority classes. To this end, we propose a novel Prior-free Balanced Replay (PBR) framework to learn from long-tailed data stream with less forgetting. Concretely, motivated by our experimental finding that the minority classes are more likely to be forgotten due to the higher uncertainty, we newly design an uncertainty-guided reservoir sampling strategy to prioritize rehearsing minority data without using any prior information, which is based on the mutual dependence between the model and samples. Additionally, we incorporate two prior-free components to further reduce the forgetting issue: (1) Boundary constraint is to preserve uncertain boundary supporting samples for continually re-estimating task boundaries. (2) Prototype constraint is to maintain the consistency of learned class prototypes along with training. Our approach is evaluated on three standard long-tailed benchmarks, demonstrating superior performance to existing CL methods and previous SOTA LTCL approach in both task- and class-incremental learning settings, as well as ordered- and shuffled-LTCL settings.

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。