L-GTA: Latent Generative Modeling for Time Series Augmentation

TL;DR

L-GTA introduces a transformer-based variational autoencoder for time series data augmentation, enabling controlled, diverse, and realistic synthetic data generation that improves predictive performance across various tasks.

Contribution

The paper proposes a novel latent generative model, L-GTA, that uses controlled transformations in the latent space to generate diverse, realistic, and task-relevant synthetic time series data.

Findings

L-GTA produces more reliable and consistent augmented data.

Synthetic data from L-GTA improves predictive accuracy.

L-GTA allows complex transformations for realistic data augmentation.

Abstract

Data augmentation is gaining importance across various aspects of time series analysis, from forecasting to classification and anomaly detection tasks. We introduce the Latent Generative Transformer Augmentation (L-GTA) model, a generative approach using a transformer-based variational recurrent autoencoder. This model uses controlled transformations within the latent space of the model to generate new time series that preserve the intrinsic properties of the original dataset. L-GTA enables the application of diverse transformations, ranging from simple jittering to magnitude warping, and combining these basic transformations to generate more complex synthetic time series datasets. Our evaluation of several real-world datasets demonstrates the ability of L-GTA to produce more reliable, consistent, and controllable augmented data. This translates into significant improvements in predictive accuracy and similarity measures compared to direct transformation methods.