Conference
Open access
Jul 2026
Anomaly Detection in Multivariate Industrial Signals: LLMs, TSFMs, or Classical Deep Learning
This paper defines a set of guiding principles (such as avoiding pre-learning or representation learning on the signals) to ensure the LLMs remain general-purpose models and proposes several algorithmic approaches for building multivariate anomaly detection pipelines.
Allen Baranov, Sarah Alnegheimish, Alfredo Cuesta-Infante et al.
· PHM Society European Confere... · 0 citations