Autoencoders for CMS tracker data quality monitoring

Data quality monitoring is an essential element of operating the CMS tracking detectors. Physicists study incoming data in real time to certify tracker data quality. Updates for Run 3 include new visualizations of tracker hit efficiency versus time and collision rate. The certification procedure could also be improved by using deep learning to automate decision making, allowing CMS to assess data in smaller time segments than human reviewers can process. We will share explorations of autoencoder neural networks for automating data quality monitoring.