Multivariate Time Series Anomaly Detection via Dual-Branch Reconstruction and Autoregressive Flow-based Residual Density Estimation
Jun Liu, Ying Chen, Ziqian Lu, Qinyue Tong, Jun Tang · 2026
Read Multivariate Time Series Anomaly Detection via Dual-Branch Reconstruction and Autoregressive Flow-based Residual Density Estimation by Jun Liu, Ying Chen, Ziqian Lu, Qinyue Tong, Jun Tang (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-08582.txt.
Description
Liu and four colleagues improve anomaly detection in multivariate time series, where summing reconstruction errors confuses hard normal cases with true anomalies. Their DBR-AF framework splits cross variable correlation learning from within variable statistics, then models leftover errors with stacked reversible flows and density estimation. Across seven benchmarks it leads rival methods, and ablations confirm both the dual branch design and the flow module matter.
Catalog record
| CREATOR | Jun Liu, Ying Chen, Ziqian Lu, Qinyue Tong, Jun Tang |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 54296 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.08582