Preserving Temporal Dynamics in Time Series Generation

2026-04-29

AI Summary

Existing time series generation methods often neglect the intrinsic temporal dynamics of the original data, leading to distributional shifts and temporal drift that degrade generation fidelity. This work proposes a model-agnostic Markov chain Monte Carlo (MCMC) framework that explicitly preserves empirical transition statistics between consecutive time points during generation, effectively correcting the error accumulation inherent in conditional generative models. As the first study to integrate MCMC into time series generation, it elucidates the mechanism of bias propagation in sequential synthesis and establishes transition law consistency as a core principle for enhancing generation quality. Evaluated on diverse datasets—including Lorenz, Licor, ETTh, and ILI—the proposed method significantly improves multiple metrics such as autocorrelation alignment, kurtosis/skewness errors, R² scores, discriminative scores, and predictive performance.

Abstract

Time-series data augmentation plays a crucial role in regression-oriented forecasting tasks, where limited data restricts the performance of deep learning models. While Generative Adversarial Networks (GANs) have shown promise in synthetic time-series generation, existing approaches primarily focus on matching marginal data distributions and often overlook the temporal dynamics that naturally exist in the original multivariate time series. When generating multivariate time series, this mismatch leads to distribution shift and temporal drift, thereby degrading the fidelity of the synthetic sequences. In this work, we propose a model-agnostic Markov Chain Monte Carlo (MCMC)-based framework to mitigate distribution shift and preserve temporal dynamics in synthetic time series. We provide a theoretical analysis of how conditional generative models accumulate deviations under sequential generation and demonstrate that the MCMC algorithm can correct these discrepancies by enforcing consistency with empirical transition statistics between neighboring time points. Extensive experiments on the Lorenz, Licor, ETTh, and ILI datasets using RCGAN, GCWGAN, TimeGAN, SigCWGAN, and AECGAN demonstrate that the proposed MCMC framework consistently improves autocorrelation alignment, skewness error, kurtosis error, R$^2$, discriminative score, and predictive score. These results suggest that synthetic time series consistent with the original data require explicit preservation of transition laws rather than solely relying on adversarial distribution matching, thereby offering a principled direction for improving generative modeling of time-series data.

Problem

• temporal dynamics
• time series generation
• distribution shift
• temporal drift
• synthetic time series

Innovation

• MCMC
• temporal dynamics
• time-series generation
• distribution shift
• transition statistics