Dynamic Linear Coregionalization for Realistic Synthetic Multivariate Time Series
Not on the map. Why?
Abstract (arXiv)
Synthetic data is essential for training foundation models for time series (FMTS), but most generators assume static correlations, and are typically missing realistic inter-channel dependencies. We introduce DynLMC, a Dynamic Linear Model of Coregionalization, that incorporates time-varying, regime-switching correlations and cross-channel lag structures. Our approach produces synthetic multivariate time series with correlation dynamics that closely resemble real data. Fine-tuning three foundational models on DynLMC-generated data yields consistent zero-shot forecasting improvements across nine benchmarks. Our results demonstrate that modeling dynamic inter-channel correlations enhances FMTS transferability, highlighting the importance of data-centric pretraining.
Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 10 of its 18 references that are in the map. First posted on arXiv 6 Apr 2026.
Builds on
10 of 18 references.
References · 10
- 2020Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecastingcited by 2,141
- 2023iTransformer: Inverted Transformers Are Effective for Time Series Forecastingcited by 906
- 2017Modeling Long- and Short-Term Temporal Patterns with Deep Neural Networkscited by 869
- 2024Chronos: Learning the Language of Time Seriescited by 712
- 2023A decoder-only foundation model for time-series forecastingcited by 650
- 2024Unified Training of Universal Time Series Forecasting Transformerscited by 588
- 2024Tiny Time Mixers (TTMs): Fast Pre-trained Models for Enhanced Zero/Few-Shot Forecasting of Multivariate Time Seriescited by 120
- 2024Toto: Time Series Optimized Transformer for Observabilitycited by 44
- 2025TimePFN: Effective Multivariate Time Series Forecasting with Synthetic Datacited by 27
- 2023A supervised generative optimization approach for tabular datacited by 1
Built on by
Not cited yet.