Using Machine Learning to Differentiate Signals from Background Events for SeaQuest

SeaQuest/E906 at Fermi National Accelerator Laboratory studies the anti-quark structure of the nucleon. A 120 GeV proton beam was collided with a series of fixed targets and created dimuons through the Drell-Yan process. The detectors observing this also observed many background muons from unrelated particle decays. We are using machine learning techniques to better separate such background events from signal events with the TMVA (Toolkit for Multivariate Analysis) package included in ROOT which contains machine learning classifiers such as Neural Networks, Boosted Decision Trees, and more. The signal samples are dimuons generated by a Monte Carlo simulation while the background samples are from real data collected via a single muon trigger. We are also exploring what kinematic variables create the most efficient markers for classification. This presentation will cover the selection of training and testing data, classifiers used, input features, and parameters that optimize the dimuon event selection for physics understanding.