A Deep Learning approach to LHCb Calorimeter reconstruction using a Cellular Automaton

On May 18, the 25th International Conference on Computing in High Energy & Nuclear Physics was held, in which the doctoral student Nuria Valls Canudas participated together with her tutors Dr. Miriam Calvo and Prof. Dr. Xavier Vilasís, where she explained her proposed alternative reconstruction algorithm for the LHCb electromagnetic calorimeter. For CERN (and in particular the LHCb experiment) the optimization of the reconstruction algorithms has become a key aspect, because it is currently undergoing a major update that will significantly increase the speed of data processing. Together with the use of deep learning techniques and an understanding of the current algorithm, the candidate's proposal is to decompose the reconstruction process into small parts that benefit the generalized learning of small neural network architectures and simplify the training data set. This approach takes the complete simulation data from the calorimeter as input and generates a list of reconstructed groups at almost constant time without any dependence on the complexity of the event.

See the conference: https://cds.cern.ch/record/2766981