发表机构
East China Normal University; Huawei Technologies Ltd.（华东师范大学; 华为技术有限公司）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
QiYao-I提出基于流形的时间注意力与频率感知动态变量交互机制，用于不规则多元时间序列预测，在基准上超越现有模型，具备强零样本和少样本泛化能力。
AI 中文摘要
不规则多元时间序列预测是现实世界应用中一个具有挑战性且重要的问题，其中观测通常是不规则采样的，并且各变量之间异步记录。现有的时间序列基础模型大多建立在规则采样的序列上，这使得它们难以泛化到不规则时间间隔和异步跨变量依赖关系。为了解决这些挑战，我们提出了QiYao-I，一种基于流形的不规则多元时间序列预测基础模型。具体来说，我们引入了一种新颖的采样条件时间流形注意力机制，该机制将真实时间戳映射到可学习的时间流形特征空间，并将时间流形偏差注入注意力层，使模型能够捕捉不规则时间间隔和局部采样结构。此外，我们提出了一种具有频率感知的动态变量交互机制。它在异步观测下选择性地执行跨变量消息传递。在真实世界不规则多元预测基准上的大量实验表明，QiYao-I在时间序列基础模型和端到端不规则预测模型方面均取得了优越的性能，在零样本和少样本设置中显示出强大的泛化能力。
英文摘要
Irregular multivariate time series forecasting is a challenging yet important problem in real-world applications, where observations are often irregularly sampled and asynchronously recorded across variables. Existing time series foundation models are mostly built on regularly sampled sequences, making them difficult to generalize to irregular time intervals and asynchronous cross-variable dependencies. To address these challenges, we propose QiYao-I, a manifold based foundation model for irregular multivariate time series forecasting. Specifically, we introduce a novel sampling-conditioned temporal manifold attention mechanism that maps real timestamps into a learnable temporal manifold feature space and injects temporal manifold biases into attention layers, enabling the model to capture both irregular time intervals and local sampling structures. Further, we propose a dynamic variable interaction mechanism with frequency awareness. It selectively performs cross-variable message passing under asynchronous observations. Extensive experiments on real-world irregular multivariate forecasting benchmarks demonstrate that QiYao-I achieves superior performance compared with both time series foundation models and end-to-end irregular forecasting models, showing strong generalization ability in zero-shot and few-shot settings.
Comments29 pages, 5 figures, 20 tables. Preprint
