用上下文学习统一时间序列预测与补全，支持不完整观测数据
TS-ICL: A Flexible Time-Indexed Foundation Model for Time Series via In-Context Learning
- 基于上下文学习的编码器-回归器架构，联合处理预测与缺失值补全
- 在补全任务上达到新最好性能，部分观测下预测精度显著提升
- 适合需要处理不规则、不完整时间序列的实际场景，如医疗监测
基础模型标志着时间序列建模的根本性转变，任务特定模型正被通用零样本模型取代。然而，现有方法主要关注预测，而真实世界的时间序列常为不规则且部分观测，需同时具备预测、填补缺失值和应对采样退化的能力。为此，我们提出TS-ICL，一种新颖的概率性上下文学习编码器-回归器Transformer，统一处理预测与补全。TS-ICL将时间序列任务建模为时间戳对齐的回归，并通过在新提出的因果数据先验生成的合成依赖结构上训练，自然引入协变量。实验表明，TS-ICL在补全任务上达到新最佳表现，同时在单变量和含协变量基准上保持与领先预测基础模型相当的竞争力。尤其在部分观测的历史窗口下表现出色。
原文摘要 · Abstract (English)
Foundation models mark a profound paradigm shift in time series modeling, with task-specific models being superseded by general-purpose zero-shot models. Yet, current approaches primarily focus on forecasting, while real-world time series are often irregularly and partially observed, requiring models that can jointly forecast, impute missing values, and handle degraded sampling conditions. To address these challenges, we introduce TS-ICL, a novel probabilistic In-Context Learning encoder--regressor Transformer that unifies forecasting and imputation. TS-ICL formulates time series tasks as timestamp-aligned regression and naturally incorporates covariates by training on synthetic dependency structures generated from a novel causal data prior. Empirically, TS-ICL achieves a new state-of-the-art in imputation, while remaining competitive with leading forecasting foundation models across both univariate and covariate-aware benchmarks. It shows particularly strong performance in forecasting with partially observed look-back windows.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。