Application of neural networks to Higgs boson search

Abstract

Abstract This paper describes an application of a neural network (NN) approach to standard model and minimal supersymmetry standard model Higgs search in the associated production t t H with H → b b . This decay channel is considered as a discovery channel for Higgs scenarios for Higgs boson masses in the range 80– 130 GeV . An NN model with a special type of data flow is used to separate t t jj background from H → b b events. This NN combines a classical NN and linear decision tree. Parameters of these NNs are randomly generated and a population of predefined size of those networks is trained as an initial generation for a subsequent genetic algorithm optimization. A genetic algorithm tunes parameters of further NN individuals derived from the previous NNs by GA operations of crossover and mutation. The goal of this GA process is optimization of the final NN performance. Our results show that NN approach is applicable to the problem of Higgs boson detection. NN filters can be used to emphasize the difference of the Mbb distribution for events accepted by filter (with better signal/background rate) and the Mbb distribution for original events (with original signal/background rate) with no loss of significance. This improvement of the shape of the Mbb distribution can be used as a criterion for existence of Higgs boson decay in considered discovery channel.