Speaker
Description
For cutting-edge detectors like CMS, data quality monitoring (DQM) and data certification (DC) are crucial components in ensuring reliable outcomes of high-level physics analysis. DQM system produces a huge number of histograms to certify, this demands a lot of human power and occasional involuntary human errors. At CMS, the present method for the DQM JetMET group is mostly reliant on manually monitoring of reference histograms summarizing the status and performance of the detector. In this talk, we present machine learning methods for certifying offline DQM data, focusing on the JetMET objects. Using DQM histograms from collision data, we show that autoencoder techniques can accurately certify data and detect ineffective detector regions in terms of JetMET, in a time period previously inaccessible for the human certification procedure.
