Machine Learning for Anomalies Detection in Large Time Series (poster) (S4D, Caen, 18/06/2018-22/06/2018)
Clément Lejeune · 2020
Context : Detect possible unexpected behaviors (anomalies) in huge amount of data recorded by sensor systems. Time series data are : •generated by complex physical phenomenons •required to be analyzed by experts-domain Goals : Detect anomalies in high-dimensional time-series. Enable realistic interpretation for experts-domain. Challenges : Learning interpretable features, coping with varying length multivariate time series, characterizing anomalies.