Exploring Deep Computing in CMS for Automated Data Validation in DQM

This project has explored the possibility of inclusion of a variational autoencoder in Automated Data Validation in DQM. The analysis has been carried out only with muon features. The main goal is to reconstruct the given lumisections and check if they can be separated between good and bad lumisections by means of the latent space representation given by the developed autoencoder. At the end, many features of good lumisections seem to be correctly reconstructed but the latent space representation does not give a proper distintion between both types of samples.