来源：时空探索之旅

文章链接：https://mp\.weixin\.qq\.com/s/6cVtPPU\_ZGpjusW7xww26Q

ACM MM 2026将在2026年11月10日至14日于巴西里约热内卢（Rio de Janeiro, Brazil）举行。本文总结了ACM MM 2026上关于时间序列的相关论文。

时间序列Topic：预测，异常检测，分类，分析，以及多模态大模型，Agentic的应用等。如有疏漏，欢迎补充！

ACM MM 2026 Programme：https://2026\.acmmm\.org/site/technical\-programme\.html

1 Closing the Loop: A Control-Theoretic Framework for Time Series Forecasting with LLMs

arXiv：https://arxiv\.org/abs/2602\.12756

代码：https://github\.com/Zh\-XY22/F\-LLM

作者：Xingyu Zhang (Institute of Software Chinese Academy of Sciences), Jingyao Wang (Institute of Software Chinese Academy of Sciences), Zeen Song (Guizhou Power Grid Co., Ltd.), Changwen Zheng (Institute of Software Chinese Academy of Sciences), Wenwen Qiang (Institute of Software Chinese Academy of Sciences)

关键词：时间序列预测、大语言模型、控制理论、闭环反馈、误差累积

2 When Should Forecasting Models Listen to Text? Counterfactual Utility Routing for Multimodal Time Series Forecasting

作者：Yusen Liu (University of Technology Sydney), Xiufeng Liu (Technical University of Denmark), Xu Cheng (Tianjin University of Technology), Huan Huo (University of Technology Sydney)

关键词：多模态时序预测，反事实

3 GeTS: Geography-aware Transformer for Multivariate Time Series Forecasting in Earth System

作者：Yuxuan Wang (Tsinghua University), Haixu Wu (Tsinghua University), Jiafeng Lin (Tsinghua University), Huakun Luo (Tsinghua University), Shiyu Wang (Independent Researcher), Zhou Ye (Independent Researcher), Jianmin Wang (Tsinghua University), Mingsheng Long (Tsinghua University)

关键词：预测，地球系统，地理感知，视觉数据

4 Empowering VLMs for Few-Shot Multimodal Time Series Classification via Tailored Agentic Reasoning

arXiv：http://arxiv\.org/abs/2605\.09395v3

代码：https://github\.com/HuangJW0821/MarsTSC

作者：Lin Li (Sun Yat-Sen University), Jiawei Huang (Sun Yat-Sen University), Qihao Quan (Sun Yat-Sen University), Dan Li (Sun Yat-Sen University), Boxin Li (Xiaomi Corporation), Xiao Zhang (Xiaomi Corporation), Erli Meng (Xiaomi Corporation), Wenjie Feng (University of Science and Technology of China), Jian Lou (Sun Yat-Sen University), See-Kiong Ng (National University of Singapore)

关键词：多模态时间序列、视觉语言模型（VLM）、智能体推理

5 TimesBook: Cross-Modal Visual Augmentation for Few-Shot Multivariate Time Series Classification

代码：https://github\.com/JJPQ/TimesBook

作者：Junjie Wang (Nanjing university), Xiangkai Ma (Nanjing university), Mingkai Lin (Nanjing university), Haoming Li (Nanjing university), Sanglu Lu (Nanjing University), Wenzhong Li (Nanjing University)

关键词：分类，少样本，跨模态，视觉增强

6 Geometry Matters: Angular Representations on the Hypersphere for Time Series Anomaly Detection

作者：Gengsen Li (Northwestern Polytechnical University), Yongjun Li (Northwestern Polytechnical University), Tonglu Li (Northwestern Polytechnical University), Xiaowen Ma (Northwestern Polytechnical University)

关键词：异常检测，超球面几何

7 GenTS: Unified Generative Modeling for Multimodal Time Series Analysis

作者：Shuqi Gu (ShanghaiTech University), Yongxiang Zhao (ShanghaiTech University), Baoyu Jing (Meta), Kan Ren (ShanghaiTech University)

关键词：多模态时序分析，生成式建模

原文链接: https://www.leiphone.com/category/private/6hAyrZHjQhvgcwbG.html