Time Series Anomaly Detection using Diffusion-based Models
This work addresses anomaly detection for time series data, but it is incremental as it applies existing diffusion methods to a new domain.
The paper tackled anomaly detection in multivariate time series by testing diffusion-based models against neural baselines, finding they outperform on synthetic datasets and are competitive on real-world ones.
Diffusion models have been recently used for anomaly detection (AD) in images. In this paper we investigate whether they can also be leveraged for AD on multivariate time series (MTS). We test two diffusion-based models and compare them to several strong neural baselines. We also extend the PA%K protocol, by computing a ROCK-AUC metric, which is agnostic to both the detection threshold and the ratio K of correctly detected points. Our models outperform the baselines on synthetic datasets and are competitive on real-world datasets, illustrating the potential of diffusion-based methods for AD in multivariate time series.