TiWeaver: Unified Temporal Dynamics Modeling via Contextual Patching

2026-06-02

AI Summary

Multivariate time series often exhibit asynchronous, irregular sampling and missing values, which challenge conventional fixed-window models in capturing complex dynamics effectively. To address this, this work proposes TiWeaver, a unified framework that introduces a Graph-guided Adaptive Tokenizer (G²AT) to enable context-consistent dynamic segmentation and designs a Fine-grained Asynchronous Dependency Extractor (FADE) to jointly capture cross-variable asynchronous relationships and long-range dependencies. Evaluated on twelve real-world datasets, the proposed method significantly outperforms existing approaches, achieving performance gains of up to 25%, and demonstrates exceptional robustness and generalization across diverse domains.

Abstract

Multivariate time series forecasting plays a critical role in real-world applications, including weather prediction, stock analysis, and health monitoring. Due to the diversity of data sources, time series exhibit diverse temporal dynamics, often accompanied by various irregularities such as missing values and non-uniform sampling frequencies. Such irregularities lead to complex and asynchronous temporal dependencies across channels. Thus, a single model with a fixed patching scheme often fails to adapt well to diverse multivariate time series, hindering accurate forecasting. In this paper, we propose TiWeaver, a unified framework designed to handle temporal dynamics and fine-grained inter-channel dependencies adaptively. Specifically, we introduce a Graph-Guided Adaptive Tokenizer (G$^2$AT) that divides time series into high contextually coherent patches by jointly considering temporal density and representation consistency. In addition, we propose a Fine-grained Asynchronous Dependency Extractor (FADE), which is designed to model fine-grained asynchronous inter-channel dependencies while incorporating long-term historical dependencies. We evaluate TiWeaver on 12 real-world time series datasets, where it achieves state-of-the-art performance, outperforming existing methods up to 25%. These results demonstrate its robustness and effectiveness across diverse domains and data characteristics.

Problem

• multivariate time series forecasting
• temporal dynamics
• irregular time series
• asynchronous dependencies
• missing values

Innovation

• adaptive tokenization
• asynchronous dependencies
• multivariate time series forecasting
• graph-guided modeling
• contextual patching