Machine Learning applications for Data Quality Monitoring and Data Certification within CMS

Abstract

Abstract The Compact Muon Solenoid (CMS) detector is getting ready for datataking in 2022, after a long shutdown period. LHC Run-3 is expected to deliver an ever-increasing amount of data. To ensure that the recorded data has the best quality possible, the CMS Collaboration has dedicated Data Quality Monitoring (DQM) and Data Certification (DC) working groups. These working groups are made of human shifters and experts who carefully watch and investigate histograms generated from different parts of the detector. However, the current workflow is not granular enough and prone to human errors. On the other hand, several techniques in Machine Learning (ML) can be designed to learn from large collections of data and make predictions for the data quality at an unprecedented speed and granularity. Hence, the data certification process can be considered as a perfect problem for ML techniques to tackle. With the help of ML, we can increase the granularity and speed of the DQM workflow and assist the human shifters and experts in detecting anomalies during data-taking. In this presentation, we present preliminary results from incorporating ML to highly granular DQM information for data certification.