Mitchel A. Perez
End-to-end Anomaly Detection, Correction and Prediction of Missing Values in Historical Daily Temperature Time Series
We present a deep learning solution to detect and correct anomalous values present in historical temperature time series, that are likely associated to human and weather instruments errors. Our solution consists in a joint peaks detection and end-to-end sequence prediction involving synchronous measurements of individual meteorological stations along with their neighboring peers. Our method was applied to temperature records of 24 meteorological stations in Belgium, with potential extensions to precipitation.
The author received his B.Sc. degree (summa cum laude) in Computer Science from the Central University of Las Villas (UCLV), and his M.Sc.Eng. degree from UCLV, Cuba, in 2011. He obtained a PhD degree in Engineering Sciences at the AVSP-ETRO Lab of the VUB in 2018. His research interests include image and video analysis, large-scale machine learning, variational models and optimization.