A novel method to measure the time-varying global association between two clusters of time series

OracleAD detects multivariate time series anomalies via prediction error and deviation from emergent latent structure, enabling interpretable, fine-grained, and root-cause diagnosis.

A dataset and benchmark for forecasting real-world time series with cause-driven irregularities and multimodal observations.

This paper introduces a novel module, IGAD, which can be integrated with reconstruction-based methods to address over generation and performance balance issues from a manifold perspective.