重新审视原型重放用于无样本持续学习：基于流形感知边界采样与自适应类别平衡损失
Revisiting Prototype Rehearsal for Exemplar-Free Continual Learning: Manifold-Aware Boundary Sampling with Adaptive Class-Balanced Loss
AI总结：
针对无样本类增量学习，提出流形感知边界采样和自适应类别平衡损失，通过生成边界感知重放样本和动态调整类别权重，使原型重放方法恢复竞争力并达到最先进性能。
摘要：
无样本类增量学习旨在随时间获取新类别而不存储原始数据。历史上，原型重放（在存储的类原型周围采样并与当前任务数据混合）是减少灾难性遗忘的流行策略。然而，最近的漂移补偿方法通过在演化特征空间中显式重新对齐原型，持续优于基于原型的重放，引发了对重放本身是否根本受限的疑问。我们认为性能差距并非源于原型重放的思想本身，而是源于其典型的实现方式：现有方法将原型视为孤立的类摘要，忽略了来自邻近敌对类的信息，并且未能纠正少量合成旧类样本与来自新引入类别的数百个真实实例之间出现的类别不平衡。基于这一假设，我们重新审视原型重放，并提出一种流形感知变体，以恢复其在无样本类增量学习中的竞争力。首先，我们引入约束扩展过采样，将每个旧类原型向其最近的新类敌对特征进行插值，生成边界感知的重放样本，这些样本更好地遵循底层数据流形，同时保持类间分离。其次，我们设计了一种自适应类别平衡损失，执行基于时间的类别加权，在旧原型信息量最大时放大其梯度，并随着后续任务积累更丰富的监督而逐渐退火其影响。这些组件共同将原型重放转变为一种抗漂移、感知不平衡的机制，缩小甚至逆转了与近期漂移补偿方法的差距，在多个无样本类增量学习基准上实现了最先进的性能。
英文摘要：
Exemplar-free class-incremental learning (EFCIL) aims to acquire new classes over time without storing raw data. Historically, prototype rehearsal, which samples around stored class prototypes and mixes them with current-task data, has been a popular strategy to reduce catastrophic forgetting. However, recent drift-compensation methods that explicitly realign prototypes in the evolving feature space consistently outperform prototype-based rehearsal, raising the question of whether rehearsal itself is fundamentally limited. We argue that the performance gap stems not from the idea of prototype rehearsal per se, but from how it is typically instantiated: existing approaches treat prototypes as isolated class summaries that ignore information from nearby enemy classes, and fail to correct the emerging class imbalance between a handful of synthetic old-class samples and hundreds of real instances from newly introduced classes. Building on this hypothesis, we revisit prototype rehearsal and propose a manifold-aware variant that restores its competitiveness in EFCIL. First, we introduce Constrained Expansive Over-Sampling, which interpolates each old-class prototype toward its nearest enemy features from new classes, generating boundary-aware rehearsal samples that better follow the underlying data manifold while preserving inter-class separation. Second, we design an Adaptive Class-Balanced loss that performs time-based class weighting, amplifying gradients from older prototypes when they are most informative and gradually annealing their influence as richer supervision from later tasks accumulates. Together, these components turn prototype rehearsal into a drift-resilient, imbalance-aware mechanism that closes, and often reverses, the gap to recent drift-compensation methods, achieving state-of-the-art performance across multiple EFCIL benchmarks.
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。
