MtsCID combines frequency-domain transformers, multi-scale patch attention, and fixed sinusoidal prototypes to detect anomalies, outperforming nine baselines on six of seven datasets.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Multivariate Time Series Anomaly Detection by Capturing Coarse-Grained Intra- and Inter-Variate Dependencies
MtsCID combines frequency-domain transformers, multi-scale patch attention, and fixed sinusoidal prototypes to detect anomalies, outperforming nine baselines on six of seven datasets.