AdaPrior：面向长尾持续学习的贝叶斯启发式自适应先验校正
AdaPrior: Bayesian-Inspired Adaptive Prior Correction for Long-Tailed Continual Learning
把长尾类增量学习重新表述为动态先验失配问题，用指数滑动平均在线估计模型诱导先验，并同时用于训练期去偏（AdaPrior Loss）与推理期轻量 logit 调整；做增量学习或长尾分类的视觉工作可直接复用这套无需额外阶段、无需样本回放的先验校正机制。
原文依据 estimate model-induced priors online via an exponential moving average and use them for (i) debiasing during training
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论长尾类增量学习（LTCIL）同时结合了两个基本挑战：对过往任务的灾难性遗忘与严重的类别不平衡。现有方法一次只缓解其中一个挑战，手段包括样本回放、重加权或分类器对齐，但它们通常假设先验是静态的，并依赖多阶段训练。与之不同，我们提出 AdaPrior，一个简单的贝叶斯框架，将 LTCIL 视为动态先验失配问题。我们的核心思想是通过指数滑动平均在线估计模型诱导的先验，并将其用于（i）训练过程中的去偏（AdaPrior 损失），以及（ii）推理时的轻量后处理校正。该组合方法统一了损失层面与推理层面的去偏，且无需额外阶段或繁重计算。我们给出理论分析，表明 AdaPrior 的先验估计器收敛到真实的模型先验，且其 logit 调整在温和假设下产生良好校准的后验。在 CIFAR100-LT、Food-101-LT、ImageNet-LT-subset 与 iNaturalist18-subset 上的大量实验表明，相较近期 LTCIL 基线取得了一致提升。除准确率之外，AdaPrior 还改善了校准与遗忘曲线，使其成为长尾持续学习中实用且可扩展的解决方案。
英文原文摘要 1281 字符
Long-Tail Class Incremental Learning (LTCIL) combines two fundamental challenges: catastrophic forgetting of past tasks and severe class imbalance. Existing approaches mitigate one challenge at a time, through rehearsal, reweighting, or classifier alignment, but they typically assume static priors and rely on multi-stage training. In contrast, we propose AdaPrior, a simple Bayesian framework that treats LTCIL as a problem of dynamic prior misalignment. Our key idea is to estimate model-induced priors online via an exponential moving average and use them for (i) debiasing during training (AdaPrior Loss), and (ii) lightweight post-hoc correction at inference. The combined approach unifies loss-level and inference-level debiasing without additional stages or heavy computation. We provide theoretical analysis showing that AdaPrior's prior estimator converges to the true model prior and that its logit adjustment yields well calibrated posteriors under mild assumptions. Extensive experiments on CIFAR100-LT, Food-101-LT, ImageNet-LT-subset, and iNaturalist18-subset demonstrate consistent gains over recent LTCIL baselines. Beyond accuracy, AdaPrior improves calibration, and forgetting curves, making it a practical and scalable solution for long-tail continual learning.