Diboson search and multivariate tools in the pp̄ → lν + heavy flavor channel at CDF

This paper describes the application of machine learning techniques to the diboson search in the lepton plus neutrino plus heavy flavor jets channel at CDF. Three different aspects of this challenging search are analyzed: multijet background rejection with the use of a support vector machine discriminant, light/heavy flavor jets separation with a 26 input variable neural network and b-jet specific energy corrections, where a resolution improvement is obtained feeding a neural network with both calorimeter and tracking information. © Società Italiana di Fisica.