发表机构
East China Normal University; Peking University; The Chinese University of Hong Kong; Shanghai Innovation Institute; Shanghai Artificial Intelligence Laboratory; University of the Chinese Academy of Sciences（华东师范大学; 北京大学; 香港中文大学; 上海创新研究院; 上海人工智能实验室; 中国科学院大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
GRAM通过振幅记忆模块和原型图模块，从冻结时间序列基础模型的预测误差中提取系统性偏差并检索修正，实现跨数据集与模型的持续预测提升。
AI 中文摘要
时间序列基础模型（TSFMs）通过大规模跨域预训练实现零样本预测，而检索增强进一步利用历史信息提升其性能。然而，现有方法通常使用相似历史窗口的真实未来值来修正TSFM的预测，这些未来值既包含基础模型已捕获的预测性成分，也包含难以迁移的样本特定随机波动。相比之下，预测误差中反复出现的系统性模型偏差更能直接刻画冻结TSFM的失效模式，因此提供了更有价值的修正信号。然而，有效利用这种模型偏差面临两个挑战：不同数值水平的预测误差因尺度差异难以比较，且重复出现的偏差必须从受随机波动污染的预测误差中提取。为应对这些挑战，我们提出GRAM，一个面向冻结TSFM的通用检索增强框架。GRAM首先引入振幅记忆模块（AMM），该模块按振幅缩放预测误差并将其聚合成可检索的原型。随后，它采用原型图模块（PGM）对原型间关系建模，以聚合一致的偏差信息同时抑制随机波动。在线预测期间，GRAM检索并扩展与当前查询相关的原型，并生成逐水平修正以细化原始TSFM预测。在多个数据集和基础模型上的实验表明，预测性能得到持续提升。
英文摘要
Time-series foundation models (TSFMs) enable zero-shot forecasting through large-scale cross-domain pretraining, while retrieval augmentation further improves their performance by leveraging historical information. However, existing methods typically correct TSFM forecasts using the ground-truth futures of similar historical windows, which contain both predictive components already captured by the foundation model and sample-specific random fluctuation that is difficult to transfer. In contrast, recurring systematic model bias within prediction errors more directly characterizes the failure modes of a frozen TSFM and therefore provides more valuable correction signals. Effectively exploiting such model bias, however, poses two challenges: prediction errors at different numerical levels are difficult to compare due to scale differences, and the recurring bias must be extracted from prediction errors contaminated by random fluctuation. To address these challenges, we propose GRAM, a general retrieval-augmented framework for frozen TSFMs. GRAM first introduces an Amplitude Memory Module (AMM) that scales prediction errors by amplitude and aggregates them into retrievable prototypes. It then employs a Prototype Graph Module (PGM) to model relations among prototypes to aggregate consistent bias information while suppressing random fluctuation. During online forecasting, GRAM retrieves and expands prototypes relevant to the current query and generates per-horizon corrections to refine the original TSFM forecast. Experiments across multiple datasets and foundation models demonstrate consistent forecasting improvements.
