IEEE ICDM 2026 Workshop

LM4TED: Large Models for Time Series Data Mining

November 15, 2026 · Shenyang, China

WELCOME TO LM4TED

Time series data are widely and continuously generated in applications including IoT systems, transportation, healthcare monitoring, and financial markets. This creates a strong demand for data mining methods that can capture temporal dependencies and support key tasks such as forecasting, anomaly detection, clustering, and classification. Large models offer a promising new paradigm for time series data mining, with the potential to enable more general-purpose representation learning and unified support for diverse downstream tasks. However, developing large models for time series data mining remains fundamentally challenging. These challenges include how to design architectures that capture complex temporal dynamics, how to achieve efficient pretraining and adaptation under limited labels and resources, and how to establish systematic evaluation across heterogeneous tasks, domains, and data distributions. In addition, real-world deployment introduces additional challenges in trustworthiness, explainability, safety, and reliability.

The goal of this workshop is to bring together researchers and practitioners working on time series data mining and large models, to better understand the opportunities and challenges in this area, to present new algorithms, systems, and evaluation methods, and to discuss applications of large models for time series data mining.

✎

CALL FOR PAPERS

We invite original research and position papers, as well as experimental studies,
on large models for time series data mining, including (but not limited to):

Large and pre-trained models for time series
Multimodal large models for time series
General-purpose representation learning for time series
Time series forecasting, anomaly detection, classification, clustering, and similarity search with large models
Efficient training, adaptation, compression, and inference for large time series models

Evaluation, benchmarks, and experimental studies for large-model-based time series data mining
Trustworthy large models for time series, including robustness, fairness, explainability, privacy, and security
Applications of large models to time series in domains such as IoT, transportation, healthcare, finance, energy, and scientific monitoring systems

◈

IMPORTANT DATES

Paper Submission

~~Aug. 20, 2026~~

Sep. 1, 2026 (AoE)

Paper Notification

~~Sep. 18, 2026~~

Sep. 14, 2026 (AoE)

Camera-ready Submission

Oct. 4, 2026 (PDT)

Workshop Date

Nov. 15, 2026, 13:30–18:00

◷

PROGRAM

Nov. 15, 2026 · 13:30–18:00 · Room Rome

13:30–13:50
Opening Remarks
13:50–14:30
Invited Talk: From Small to Large Models for Time-Series Decision-Making Yushuai Li, Aalborg University
Paper Session 112 min talk + 3 min Q&A per paper
14:30–14:45
DM321When Do Foundation Models Pay Off? A Break-Even Analysis of Pretrained Time Series Forecasters
14:45–15:00
S10209Beyond Scaling: Bayesian Temporal Inductive Bias for Time-Series Foundation Model Adaptation
15:00–15:15
S10207Beyond Forecasting: Operationalizing Foundational Time Series Models for Industrial Digital Twins
15:15–15:30
DM949DTMTimer: Dual-Path Token Merging and Cross-Modal Fusion for Multivariate Time Series Forecasting with Task-Aware LLMs
15:30–15:50
Coffee Break
Paper Session 212 min talk + 3 min Q&A per paper
15:50–16:05
S10201SWAP: Adaptive Tabular In-Context Learning
16:05–16:20
S10204VibCA-Fusion: Self-supervised Contrastive Learning for Label-Efficient Industrial Vibration Sensing and Transformer Fault Detection
16:20–16:35
S10210LLM Agents for Anomaly Detection and Root-Cause Classification in Refrigeration Systems
16:50–17:05
S10205VeriTS: Verifiable Model-Enhanced Time-Series Queries on Blockchain Systems
16:35–16:50
S10206Time Series Agentic AI for Energy Systems
17:05–17:20
S10208TS-HotIndex: Time-Series-Aware Predictive Query Indexing for Blockchain Data Retrieval
17:20–17:35
Best Paper Award Ceremony
17:35–18:00
Closing Remarks & Workshop Group Photo (all participants)

✉

SUBMISSION

Submission Site: LM4TED Submission Site

Regular papers: Up to 8 pages, plus up to 2 extra pages at USD 25 per page.
Short papers: Up to 6 pages, plus up to 2 extra pages at USD 25 per page.
Submissions should follow the ICDM 2026 template.
Accepted papers will be included in the ICDM Workshop Proceedings, separate from the ICDM Main Conference Proceedings.
Each accepted workshop paper requires a full registration.
Best Paper Awards will be selected and presented at the workshop.
This workshop will follow a single-blind review process.
Duplicate submissions of the same paper to more than one ICDM workshop are forbidden.