TimeLDM: Latent Diffusion Model for Unconditional Time Series Generation

TL;DR

TimeLDM introduces a latent diffusion model utilizing a variational autoencoder for high-quality, robust, and state-of-the-art synthetic time series generation across various datasets and lengths.

Contribution

The paper presents a novel latent diffusion approach for time series generation that outperforms existing methods in quality and robustness.

Findings

Achieves state-of-the-art results on simulated benchmarks.

Improves Discriminative score by 55% on average.

Enhances Context-FID and Discriminative scores by 80% and 50%, respectively.

Abstract

Time series generation is a crucial research topic in the area of decision-making systems, which can be particularly important in domains like autonomous driving, healthcare, and, notably, robotics. Recent approaches focus on learning in the data space to model time series information. However, the data space often contains limited observations and noisy features. In this paper, we propose TimeLDM, a novel latent diffusion model for high-quality time series generation. TimeLDM is composed of a variational autoencoder that encodes time series into an informative and smoothed latent content and a latent diffusion model operating in the latent space to generate latent information. We evaluate the ability of our method to generate synthetic time series with simulated and real-world datasets and benchmark the performance against existing state-of-the-art methods. Qualitatively and quantitatively, we find that the proposed TimeLDM persistently delivers high-quality generated time series. For example, TimeLDM achieves new state-of-the-art results on the simulated benchmarks and an average improvement of 55% in Discriminative score with all benchmarks. Further studies demonstrate that our method yields more robust outcomes across various lengths of time series data generation. Especially, for the Context-FID score and Discriminative score, TimeLDM realizes significant improvements of 80% and 50%, respectively. The code will be released after publication.