Latent flow time series generation aligned to spectral features
Time series generation with spectrally aligned latent flow matching
Machine Learning
Summary
Generating realistic time series data is important but challenging because compressing data into a simpler form can distort important details. The authors improve this by training a model that preserves key signal properties like smoothness and frequency content, using tools like Fourier and wavelet transforms. This leads to synthetic time series that better match real data in behavior and structure. Tests on real-world datasets show their method creates more realistic and computationally efficient time series than previous approaches.
What this means in practice
- •For financial data analysts: Generate realistic synthetic financial time series that preserve volatility and cyclical patterns for risk modeling and strategy testing.
- •For medical monitoring teams: Create accurate synthetic physiological signals such as heart rate or EEG for training and validating diagnostic algorithms without risking patient privacy.
Authors
Camilo Carvajal Reyes, Felipe Tobar
Abstract
Latent flow models have proven to be a reliable and cost-effective method for time series generation. However, the latent compression induces unwanted artefacts, such as a spectral mismatch with respect to the underlying dataset, thus hindering their use as training surrogates. In this article, we propose a spectrally-aligned latent-flow time series generator, where the latent space for flow matching is trained to preserve dynamical properties that are relevant for the suitability of synthetic samples. We find that incorporating fine-tuning losses based on canonical signal representations such as the Fourier, wavelet and signature transforms helps overcome these issues. The interpretability of these transformations allows us to ensure that the synthetic signals are aligned with the true ones in terms of relevant features, such as smoothness or targeted spectral content, as opposed to relying on pointwise reconstruction losses only. We compare the proposed aligned models against a base latent-flow model and the state of the art over real-world long-range univariate and multivariate benchmark datasets. Our quantitative results validate the superiority of the proposed method in terms of its performance on metrics reflecting signal realness and computational efficiency, while being aligned to the training set with respect to its local structure.