VLDB 2025 | 时间序列（Time Series）论文总结（预测，异常检测，压缩，自动化等）

VLDB 2025于2025年9月1号-5号在英国伦敦(London, United Kingdom)举行。

本文总结了VLDB 2025 有关时间序列（Time Series）的相关论文，主要包含如有疏漏，欢迎大家补充。

时间序列Topic：预测，异常检测，聚类，压缩，自动化，大模型，时序数据库等。

1. Less is More: Efficient Time Series Dataset Condensation via Two-fold Modal Matching 2. A Memory Guided Transformer for Time Series Forecasting 3. Goku: A Schemaless Time Series Database for Large Scale Monitoring at Pinterest 4. Fully Automated Correlated Time Series Forecasting in Minutes 5. Discovering Leitmotifs in Multidimensional Time Series 6. MLP-Mixer based Masked Autoencoders Are Effective, Explainable and Robust for Time Series Anomaly Detection 7. Representative Time Series Discovery for Data Exploration 8. Migration-Free Elastic Storage of Time Series in Apache IoTDB 9. Streaming Time Series Subsequence Anomaly Detection: A Glance and Focus Approach 10. Unsupervised Anomaly Detection in Multivariate Time Series across Heterogeneous Domains 11. Time Series Motif Discovery: A Comprehensive Evaluation 12. ChatTS: Aligning Time Series with LLMs via Synthetic Data for Enhanced Understanding and Reasoning 13. STsCache: An Efficient Semantic Caching Scheme for Time-series Data Workloads Based on Hybrid Storage 14. TAB: Unified Benchmarking of Time Series Anomaly Detection Methods 15. UFGTime: Mining Intertwined Dependencies in Multivariate Time Series via an Efficient Pure Graph Approach (Flavor: Foundations and Algorithms Papers) 16. MOMENTI: Scalable Motif Mining in Multidimensional Time Series 17. Improving Time Series Data Compression in Apache IoTDB 18. TSB-AutoAD: Towards Automated Solutions for Time-Series Anomaly Detection

E, A \& B

19. Time-Series Clustering: A Comprehensive Study of Data Mining, Machine Learning, and Deep Learning Methods 20. Demonstration of ModelarDB: Model-Based Management of High-Frequency Time Series Across Edge, Cloud, and Client 21. EasyAD: A Demonstration of Automated Solutions for Time-Series Anomaly Detection 22. SAIL: A Voyage to Symbolic Approximation Solutions for Time-Series Analysis

🌟【紧跟前沿】"时空探索之旅"与你一起探索时空奥秘！🚀

欢迎大家关注时空探索之旅时空探索之旅

链接：++https://www.vldb.org/pvldb/vol18/p226-miao.pdf++

代码：++https://github.com/uestc-liuzq/STdistillation++

作者：Hao Miao, Ziqiao Liu, Yan Zhao, Chenjuan Guo, Bin Yang, Kai Zheng, Christian S. Jensen

关键词：时序数据压缩，模态匹配

2 A Memory Guided Transformer for Time Series Forecasting

链接：++https://www.vldb.org/pvldb/vol18/p239-cheng.pdf++

代码：++https://github.com/YunyaoCheng/Memformer++

作者：Yunyao Cheng, Chenjuan Guo, Bin Yang, Haomin Yu, Kai Zhao, Christian S. Jensen

关键词：预测，Transformer，记忆

3 Goku: A Schemaless Time Series Database for Large Scale Monitoring at Pinterest

链接：++https://www.vldb.org/pvldb/vol18/p503-sanghavi.pdf++

作者：Monil Mukesh Sanghavi, Ming-May Hu, Zhenxiao Luo, Xiao Li, Kapil Bajaj

关键词：无模式时序数据库

4 Fully Automated Correlated Time Series Forecasting in Minutes

链接：++https://www.vldb.org/pvldb/vol18/p144-wu.pdf++

代码：++https://github.com/ccloud0525/FACTS++

作者：Xinle Wu, Xingjian Wu, Dalin Zhang, Miao Zhang, Chenjuan Guo, Bin Yang, Christian S. Jensen

关键词：全自动化预测，效率

5 Discovering Leitmotifs in Multidimensional Time Series

链接：++https://www.vldb.org/pvldb/vol18/p377-schafer.pdf++

代码：++https://github.com/patrickzib/leitmotifs++

作者：Patrick Schäfer, Ulf Leser

关键词：主导动机发现,多维时间序列

6 MLP-Mixer based Masked Autoencoders Are Effective, Explainable and Robust for Time Series Anomaly Detection

链接：++https://www.vldb.org/pvldb/vol18/p798-qideng.pdf++

代码：++https://github.com/richard-tang199/MMA++

作者：Tang Qideng, Dai Chaofan, Wu Yahui, Zhou Haohao

关键词：异常检测，MAE，MLP-Mixer

7 Representative Time Series Discovery for Data Exploration

链接：++https://www.vldb.org/pvldb/vol18/p915-bao.pdf++

代码：++https://github.com/rmitbggroup/RTSD++

作者：Ge Lee, Shixun Huang, Zhifeng Bao, Yanchang Zhao

关键词：时序相似度度量

8 Migration-Free Elastic Storage of Time Series in Apache IoTDB

链接：++https://www.vldb.org/pvldb/vol18/p1784-song.pdf++

作者：Rongzhao Chen, Xiangpeng Hu, Xiangdong Huang, Chen Wang, Shaoxu Song, Jianmin Wang

关键词：弹性时序存储，IoTDB

9 Streaming Time Series Subsequence Anomaly Detection: A Glance and Focus Approach

链接：++https://www.vldb.org/pvldb/vol18/p1892-zheng.pdf++

代码：++https://github.com/Wangwenjing1996/Sirloin++

作者：Wenjing Wang, Ziyang Yue, Bolong Zheng

关键词：异常检测，流式时序

10 Unsupervised Anomaly Detection in Multivariate Time Series across Heterogeneous Domains

链接：++https://www.vldb.org/pvldb/vol18/p1691-jacob.pdf++

代码：++https://github.com/exathlonbenchmark/divad++

作者：Vincent Jacob, Yanlei Diao

关键词：异常检测，无监督，异构域

11 Time Series Motif Discovery: A Comprehensive Evaluation

链接：++https://www.vldb.org/pvldb/vol18/p2226-boniol.pdf++

代码：++https://github.com/grrvlr/TSMD++

作者：Valerio Guerrini, Thibaut Germain, Charles Truong, Laurent Oudre, Paul Boniol

关键词：模式（基序）发现

12 ChatTS: Aligning Time Series with LLMs via Synthetic Data for Enhanced Understanding and Reasoning

链接：++https://www.vldb.org/pvldb/vol18/p2385-xie.pdf++

代码：++https://github.com/NetManAIOps/ChatTS++

作者：Zhe Xie, Zeyan Li, Xiao He, Longlong Xu, Xidao Wen, Tieying Zhang, Jianjun Chen, Rui Shi, Dan Pei

关键词：LLM，时序推理，对齐，智能运维

13 STsCache: An Efficient Semantic Caching Scheme for Time-series Data Workloads Based on Hybrid Storage

链接：++https://www.vldb.org/pvldb/vol18/p2964-li.pdf++

代码：++https://github.com/ts-lab1024/ts-semantic-caching++

作者：Tao Kong, Hui Li, Yuxuan Zhao, Liping Li, Xiyue Gao, Qilong Wu, Jiangtao Cui

关键词：时间序列数据工作负载，查询模式

14 TAB: Unified Benchmarking of Time Series Anomaly Detection Methods

链接：++https://www.vldb.org/pvldb/vol18/p2775-hu.pdf++

代码：++https://github.com/decisionintelligence/TAB++

作者：Xiangfei Qiu, Zhe Li, Wanghui Qiu, Shiyan Hu, Lekui Zhou, Xingjian Wu, Zhengyu Li, Chenjuan Guo, Aoying Zhou, Zhenli Sheng, Jilin Hu, Christian S. Jensen, Bin Yang

关键词：异常检测，benchmark

15 UFGTime: Mining Intertwined Dependencies in Multivariate Time Series via an Efficient Pure Graph Approach (Flavor: Foundations and Algorithms Papers)

链接：++https://www.vldb.org/pvldb/vol18/p3175-gao.pdf++

代码：++https://github.com/WonderHeiYi/UFGTIME++

作者：Ruikun Li, Dai Shi, Ye Xiao, Junbin Gao

关键词：多元时序预测，GNN

16 MOMENTI: Scalable Motif Mining in Multidimensional Time Series

链接：++https://www.vldb.org/pvldb/vol18/p3463-ceccarello.pdf++

代码：++https://github.com/aidaLabDEI/MOMENTI-motifs++

作者：Matteo Ceccarello, Francesco Pio Monaco, Francesco Silvestri

关键词：基序挖掘，多维时序

17 Improving Time Series Data Compression in Apache IoTDB

链接：++https://www.vldb.org/pvldb/vol18/p3406-tang.pdf++

代码：++https://github.com/yuxin370/CompressIoTDB++

作者：Yuxin Tang, Feng Zhang, Jiawei Guan, Yuan Tian, Xiangdong Huang, Chen Wang, Jianmin Wang, Xiaoyong Du

关键词：时序数据压缩，模态匹配

18 TSB-AutoAD: Towards Automated Solutions for Time-Series Anomaly Detection $E, A \& B$

链接：++https://www.vldb.org/pvldb/vol18/p4364-liu.pdf++

代码：++https://github.com/TheDatumOrg/TSB-AutoAD++

作者：Qinghua Liu, Seunghak Lee, Paparrizos John

关键词：异常检测，自动化

19 Time-Series Clustering: A Comprehensive Study of Data Mining, Machine Learning, and Deep Learning Methods

链接：++https://www.vldb.org/pvldb/vol18/p4380-paparrizos.pdf++

代码：++http://www.timeseries.org/tsclusteringeval++

作者：John Paparrizos, Sai Prasanna Teja Reddy Bogireddy

关键词：时序聚类

20 Demonstration of ModelarDB: Model-Based Management of High-Frequency Time Series Across Edge, Cloud, and Client

链接：++https://www.vldb.org/pvldb/vol18/p5247-jensen.pdf++

作者：Søren Kejser Jensen, Christian Schmidt Godiksen, Christian Thomsen, Torben Bach Pedersen

关键词：边缘部署，ModelarDB

21 EasyAD: A Demonstration of Automated Solutions for Time-Series Anomaly Detection

链接：++https://www.vldb.org/pvldb/vol18/p5431-liu.pdf++

作者：Qinghua Liu, Seunghak Lee, John Paparrizos

关键词：异常检测，自动化解决方案

22 SAIL: A Voyage to Symbolic Approximation Solutions for Time-Series Analysis

链接：++https://www.vldb.org/pvldb/vol18/p5419-yang.pdf++

代码：++https://github.com/TheDatumOrg/SAIL++

作者：Fan Yang, John Paparrizos

关键词：时序分析，符号分解