Searches for new physics with Machine Learning at the LHC

In this project a classifier of new physics events is developed using machine learning, in particular events recorded at the ATLAS and CMS detectors at the LHC. A deep neural network model which is a modification of an architecture know as convDAN is used. This model well represents the the structure of jet level event data and as such outperforms hand designed variables in a number of metrics. This work provides motivation for further in- vestigation into the utility of deep neural network models in for the classification of new physics events. This work was carried out as part of a summer project in the particle physics group at the University of Bristol from 10/06/19 to 2/08/19. This work was supervised by Tai Sakuma and Henning Flaecher