arXiv cs.AI· Yangze Li, Zhengyang Zhou, Qihe Huang, Pengkun Wang, Binwu Wang, Yang Wang·· 3 天前AI 评分20
TStruct：为长期时间序列预测学习共享时间结构
TStruct: Learning Shared Temporal Structures for Long-Term Time-Series Forecasting
AI 导读
针对长期时间序列预测中噪声与非平稳性导致模型易拟合偶然时序交互的问题，研究者提出 Temporal Structure Network（TStruct），通过直接参数化一组共享时间依赖矩阵，并利用动态样本信息与静态变量先验生成归一化组合权重，实现共享结构约束下的自适应时间依赖建模。该结构感知设计在所有数据集上均优于多个强基线，同时保持较高计算效率，论文已被 NeurIPS 2026 主赛道接收。
正文 · 原文
Abstract:Long-term time series forecasting requires temporal dependency modeling that remains stable and generalizable under temporal distribution shifts. However, real-world time series are often noisy and non-stationary, making deep learning models prone to fitting incidental temporal interactions in the training data. A key question is whether there exist temporal relationship structures that recur across different temporal positions and variables. We analyze temporal relationship matrices on commonly used time series datasets and find that samples from different temporal positions and different variables exhibit shared dominant temporal relationship structures. This indicates that temporal relationship structures contain patterns that recur stably across samples and variables. To learn such shared structures, we propose \textbf{T}emporal \textbf{Struct}ure Network (\textbf{TStruct}), a simple yet effective network for directly learning shared and generalizable temporal dependencies. Specifically, TStruct directly parameterizes a set of shared temporal dependency matrices and generates normalized combination weights using dynamic sample information and static variable priors. These weights represent a soft allocation of each sample-variable pair over shared temporal dependency structures, enabling the model to generate adaptive temporal dependency matrices under shared structural constraints, rather than fitting incidental interactions in a fully unconstrained manner. Notably, this simple structure-aware design consistently outperforms several strong baselines across all datasets, achieving superior accuracy while maintaining high computational efficiency.
| Comments: | This paper has been accepted to NeurIPS 2026(Main Track) |
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.17864 [cs.LG] |
| (or arXiv:2610.17864v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.17864 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Yangze Li [view email]
[v1]
Tue, 13 Oct 2026 07:24:58 UTC (2,820 KB)
来源：arXiv cs.AI · arxiv.org