REGEN: Reference-Guided Synthetic Multivariate Time Series Generation for Forecasting

Researchers have developed ReGeN, a reference-guided generative pipeline that creates synthetic multivariate time series data by decomposing observed sequences into periodic backbones, stochastic residuals, and cross-variable dependencies. The system outperforms existing methods by producing domain-grounded samples that can substitute for real data with minimal forecasting degradation, and in traffic domains, synthetic data even surpassed real source performance. This approach addresses the critical challenge of training robust forecasting models in low-data regimes by structurally exploiting reference data rather than simply imitating it.

arXiv:2606.05264v1 Announce Type: new Abstract: Training robust multivariate time series forecasting models requires large, diverse corpora, yet many real-world domains provide only a handful of observed sequences. Existing generators fail to resolve this mismatch: prior-based approaches e.g., CauKer, TimePFN produce domain-agnostic samples, while data-driven methods e.g., TimeGAN treat references as black-box supervision, forfeiting explicit control over periodic structure, local variability, and cross-variable dynamics. We propose ReGeN, a reference-guided generative pipeline that treats observed sequences not as examples to imitate, but as structural scaffolds for controllable synthesis. ReGeN decomposes each reference into three interpretable components: a phase-aligned periodic backbone capturing dominant domain morphology; per-variable stochastic residuals modeled with a deep-kernel Gaussian process; and lag-aware cross-variable dependencies injected through a structural causal model with fitted coupling coefficients. Sampling these components at controllable temperature broadens distributional coverage while preserving domain-grounded structure. We show that ReGeN-generated data consistently substitutes for real sibling data with minimal forecasting degradation, and in strongly periodic domains such as traffic, can outperform the real source itself. We further show that a foundation model pretrained on ReGeN corpora outperforms those pretrained on prior-based and data-driven synthetic alternatives. This suggests that in low-data regimes, how reference data is structurally exploited can matter as much as how much data is available.