MulGad: Multi-granularity contrastive learning for multivariate time series anomaly detection

Open Webpage

Bo-Wen Xiao, Hong-Jie Xing, Chun-Guo Li

Published: 01 Jan 2025, Last Modified: 09 Nov 2025Inf. Fusion 2025EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•The proposed MulGad can capture both coarse-grained and fine-grained information.•λ<math><mi is="true">λ</mi></math>-contextual contrastive learning is proposed to learn fine-grained information.•The cross-reconstruction based contrastive learning is designed.•Linear attention network is used for the first time in time series analysis.